nanonyous
Expert Member
Allow me to club you over the head with your PSU. If you hold still enough, I just might be able to improve your ventilation.
(j/k, unless you're serias)
-
Join the MyBroadband community
South Africa’s biggest forum. Discuss, discover, and connect with thousands of members.
-
Question of the Day: What's the most interesting small town you've visited in South Africa?
PC performance bottlenecks identified
- Thread starter rpm
- Start date
Allow me to club you over the head with your PSU. If you hold still enough, I just might be able to improve your ventilation.
(j/k, unless you're serias)
I still remember these 5MB Apple Profile drives, one mate had one and his dad was stinking rich
nanonyous
Expert Member
Kids these days and their hard drives. Back in my day, you loaded DOS into RAM and had to reload DOS every time you exited an application after loading it from its own floppy! Having a computer that could use and even had two floppy drives in it was a damned luxury! We had to load our programs up(hill) both ways (in the) slow!
Jab
Expert Member
Floppy drives? Damn whippersnappers. In my days we used tapes to load our applications. This gave you enough time while waiting for the program to load to go make a sandwich, have swim & watch some TV (that is if it was after 4 and the TV have started)
Derrick
ლ(ಠ_ಠ )ლ
- Joined
- Nov 22, 2010
- Messages
- 5,085
- Reaction score
- 5
Post 1(too long for a single post)
Low framerates are often an indication of a laggy system. Also, when framrates dip due to the system not being able to keep up, games commonly refer to that as lag. [/B]
Now that that's out of the way...
Graphics cards alone are not responsible for low framerates when one is comparing low and high settings; if the player is trying to employ PhysX or any other implementation of physics and doesn't even realise they're enabling such a setting, then their CPU may be taxed by an unreasonable amount. Metro2033 has an excellent example of this in the OpenCL based depth-of-field setting, which still brings many high-end computers to their knees.
You misunderstand the “how to test” section of the graphics card category. This is mean as an easy way to test whether the graphics card is the bottleneck in the system. If there is no change between the lag a gamer experiences after they turn down the graphics card settings, the chances of the graphics card being the component causing the lag is low. Inversely, if turning down graphics settings addresses the lag issue, it is the graphics card.
There are other aspects to take into consideration that do not necessarily tie directly into the graphics card's raw performance capacity, and that's in the amount of graphics memory that needs to be utilized in order for textures and screen-space processing effects to work. If your card has insufficient memory but an abundance of raw power, it could run many modern games at their maximum settings provided you keep the texture setting and/or post-processing settings such as AA, HDR (which in many games is really just a form of bloom, not real HDR, anyway...), Bloom, motion blur and depth of field are kept at low settings or even turned off entirely.
This is most evident in cards with less than 768mb of ram these days as games running at 1920x1080 with all the bells and whistles can easily fill the framebuffer (RAM) of the card immediately. Windows7 and the DirectX10/11 rendering paths are able to utilize system ram as a fallback, but this is hardly ideal if your system's ram is already saturated anyway, and that doesn't even take into consideration calls from the CPU and GPU both needing to be serviced simultaneously; there simply isn't as much bandwidth left for the GPU to use. Most graphics cards in the mainstream to enthusiast tiers these days are working with over 30Gbps worth of memory bandwidth while most mainstream to high-end computers' system ram offer between 17-21Gbps total bandwidth - see how that can be a bad thing? Running your cards in SLI or crossfire with the hope that combining two 512mb or 768mb cards into 1 or 1.5gb worth of ram doesn't work, either. Each card has to clone the same information for their respective framebuffers (ram. I'll repeat it every time so that there's no room for confusion), so you are not going to be magically doubling your total graphics ram capacity. In general, the higher the resolution you want to play at, the higher your card's ram capacity ought to be to keep up.
I’m glad to see you went in depth into graphics cards but I’m not sure how this relates to what is obviously an article addressed at a lower level, with more in depth (and more technical) articles to follow. I’m curious though, since you went so in depth into the inner workings of graphics cards, why you didn’t mention Memory bus width? It ties in perfectly with your “insufficient memory but an abundance of raw power”. Or shader processors? What about Raster operators? Texture units?
A CPU is also not responsible for 'lag' on its own. See the above notes about 'graphics' settings that actually utilize the processor where the graphics card's preoccupied or simply is not responsible for/cannot be the one to handle a particular operation. These days, if you have a processor that's at least of the Core2Duo generation in the E6xxx or greater series or in the Athlon II X3 (tri-core) processor realm, your processor is almost guaranteed to not be responsible for that significant a framerate loss. What is of use however is to use any program that can perform background monitoring of your processor's active time. Task Manager alone can even perform this task for you, you simply have to open it up (ctrl+shift+esc) then head to the performance tab and let it do its merry thing in the background while you're playing your game. When you encounter a section of the game that's causing a particularly low framerate, stick around in that part of the game for a while, trying to keep the framerate low, then alt-tab out of your game (even if it'll 'break' the game) and look at the charts. Are all your 'threads' being utilized in full? Is only one being utilized very heavily? Are your threads not one being utilized in full, but all of them are seeing relatively high amounts of usage?
If only one thread out of several is full, that means the game is not threading optimized and is only utilizing one 'core'. In this case, you could try increasing the frequency of your processor to increase its overall instructions per second (IPS, see here for a brief explanation) per core. It's extremely important to note here that not all processors are equal when it comes to per-core IPS throughput and that just because a processor has 6 cores vs its 4-core version (such as the Phenom hexa-cores vs the quad-cores) does not automatically mean that it will have greater IPS per core; it can in fact have less on account of running at a lower frequency at stock settings.
If only one thread is running at less than full capacity, there's a good chance there's either a memory bandwidth issue either to system ram or to the cache on the processor, or getting information from CPU to graphics card's RAM (ie: graphics card memory bandwidth constraint); the cache capacity could be insufficient or it could simply not have the bandwidth that stream of instructions would require of it for optimum performance. See the three below links for an example of how different processors from exactly the same generation have different cache amounts. Technically, these processors are all equal, they are simply using different cache amounts and run at different stock FSB/multiplier settings.
Intel Celeron E3400 2.66ghz 1mb cache 800mhz FSB
Intel Pentium E5400 2.7ghz 2mb cache 800mhz FSB
Intel E7300 2.66ghz 3mb cache 1066mhz FSB
All three of these processors are exactly the same, they simply each have 1mb more L2 cache and use slightly different FSB/core multiplier combinations.
All cores being fully utilized? Your system RAM and graphics card(s) are likely keeping up here bandwidth-wise (not necessarily performance-wise, however), and cache is clearly not an issue. The game is making decent use of all available cores/threads, even though such games are extremely rare. At this point you're either looking at a limit in the IPS capacity of your processor or something else may be causing a framerate drop. Few (if any?) games are able to fully utilize processors at all anyway, but let's ignore that for now.
Very in depth and well written, but you try to qualify why the CPU would cause lag in a game. CPUs can cause lag in games, particularly when the game is poorly optimised and doesn’t make use of all cores available. Is the CPU the bottleneck in this situation? Yes it is, because upgrading to a faster CPU with faster single threaded performance (due to increased cache size, faster FSB speeds/lower core multipliers, more efficient, faster core speed etc) will result in less lag, removing the bottleneck. In a perfect world every game would utilise every core of every CPU, however this isn’t the case. Also, blanket statements about “if you have at least XXX processor you won’t experience CPU lag”. While this may or may not be true for the majority of games, that isn’t always the case. Have you tested every CPU/GPU combination with every modern game? Once again, the focus of the article isn’t as high level as your comment. Interestingly, you didn’t detail how the memory controller (as most modern systems memory controllers are situated in the CPU itself). Also, why would you say technically all of the above CPUs are equal? The difference extends beyond the info listed on the Intel site. You also missed out how the E7300 loses out VT support compared to the other two processors but I’ll let that slide. What I can’t let slide is how the E7300 supports the SSE4.1 instruction set whereas the E3400 doesn’t? That’s a fairly large difference in the context of your argument This is another example of the differences between CPUs beyond the obvious (cores, speed, FSB, multiplier L2 cache). To say an E3400 and E7300 is exactly the same besides L2 cache and FSB speed/clock multiplier is ignorant.
Low framerates are often an indication of a laggy system. Also, when framrates dip due to the system not being able to keep up, games commonly refer to that as lag. [/B]
Now that that's out of the way...
Graphics cards alone are not responsible for low framerates when one is comparing low and high settings; if the player is trying to employ PhysX or any other implementation of physics and doesn't even realise they're enabling such a setting, then their CPU may be taxed by an unreasonable amount. Metro2033 has an excellent example of this in the OpenCL based depth-of-field setting, which still brings many high-end computers to their knees.
You misunderstand the “how to test” section of the graphics card category. This is mean as an easy way to test whether the graphics card is the bottleneck in the system. If there is no change between the lag a gamer experiences after they turn down the graphics card settings, the chances of the graphics card being the component causing the lag is low. Inversely, if turning down graphics settings addresses the lag issue, it is the graphics card.
There are other aspects to take into consideration that do not necessarily tie directly into the graphics card's raw performance capacity, and that's in the amount of graphics memory that needs to be utilized in order for textures and screen-space processing effects to work. If your card has insufficient memory but an abundance of raw power, it could run many modern games at their maximum settings provided you keep the texture setting and/or post-processing settings such as AA, HDR (which in many games is really just a form of bloom, not real HDR, anyway...), Bloom, motion blur and depth of field are kept at low settings or even turned off entirely.
This is most evident in cards with less than 768mb of ram these days as games running at 1920x1080 with all the bells and whistles can easily fill the framebuffer (RAM) of the card immediately. Windows7 and the DirectX10/11 rendering paths are able to utilize system ram as a fallback, but this is hardly ideal if your system's ram is already saturated anyway, and that doesn't even take into consideration calls from the CPU and GPU both needing to be serviced simultaneously; there simply isn't as much bandwidth left for the GPU to use. Most graphics cards in the mainstream to enthusiast tiers these days are working with over 30Gbps worth of memory bandwidth while most mainstream to high-end computers' system ram offer between 17-21Gbps total bandwidth - see how that can be a bad thing? Running your cards in SLI or crossfire with the hope that combining two 512mb or 768mb cards into 1 or 1.5gb worth of ram doesn't work, either. Each card has to clone the same information for their respective framebuffers (ram. I'll repeat it every time so that there's no room for confusion), so you are not going to be magically doubling your total graphics ram capacity. In general, the higher the resolution you want to play at, the higher your card's ram capacity ought to be to keep up.
I’m glad to see you went in depth into graphics cards but I’m not sure how this relates to what is obviously an article addressed at a lower level, with more in depth (and more technical) articles to follow. I’m curious though, since you went so in depth into the inner workings of graphics cards, why you didn’t mention Memory bus width? It ties in perfectly with your “insufficient memory but an abundance of raw power”. Or shader processors? What about Raster operators? Texture units?
A CPU is also not responsible for 'lag' on its own. See the above notes about 'graphics' settings that actually utilize the processor where the graphics card's preoccupied or simply is not responsible for/cannot be the one to handle a particular operation. These days, if you have a processor that's at least of the Core2Duo generation in the E6xxx or greater series or in the Athlon II X3 (tri-core) processor realm, your processor is almost guaranteed to not be responsible for that significant a framerate loss. What is of use however is to use any program that can perform background monitoring of your processor's active time. Task Manager alone can even perform this task for you, you simply have to open it up (ctrl+shift+esc) then head to the performance tab and let it do its merry thing in the background while you're playing your game. When you encounter a section of the game that's causing a particularly low framerate, stick around in that part of the game for a while, trying to keep the framerate low, then alt-tab out of your game (even if it'll 'break' the game) and look at the charts. Are all your 'threads' being utilized in full? Is only one being utilized very heavily? Are your threads not one being utilized in full, but all of them are seeing relatively high amounts of usage?
If only one thread out of several is full, that means the game is not threading optimized and is only utilizing one 'core'. In this case, you could try increasing the frequency of your processor to increase its overall instructions per second (IPS, see here for a brief explanation) per core. It's extremely important to note here that not all processors are equal when it comes to per-core IPS throughput and that just because a processor has 6 cores vs its 4-core version (such as the Phenom hexa-cores vs the quad-cores) does not automatically mean that it will have greater IPS per core; it can in fact have less on account of running at a lower frequency at stock settings.
If only one thread is running at less than full capacity, there's a good chance there's either a memory bandwidth issue either to system ram or to the cache on the processor, or getting information from CPU to graphics card's RAM (ie: graphics card memory bandwidth constraint); the cache capacity could be insufficient or it could simply not have the bandwidth that stream of instructions would require of it for optimum performance. See the three below links for an example of how different processors from exactly the same generation have different cache amounts. Technically, these processors are all equal, they are simply using different cache amounts and run at different stock FSB/multiplier settings.
Intel Celeron E3400 2.66ghz 1mb cache 800mhz FSB
Intel Pentium E5400 2.7ghz 2mb cache 800mhz FSB
Intel E7300 2.66ghz 3mb cache 1066mhz FSB
All three of these processors are exactly the same, they simply each have 1mb more L2 cache and use slightly different FSB/core multiplier combinations.
All cores being fully utilized? Your system RAM and graphics card(s) are likely keeping up here bandwidth-wise (not necessarily performance-wise, however), and cache is clearly not an issue. The game is making decent use of all available cores/threads, even though such games are extremely rare. At this point you're either looking at a limit in the IPS capacity of your processor or something else may be causing a framerate drop. Few (if any?) games are able to fully utilize processors at all anyway, but let's ignore that for now.
Very in depth and well written, but you try to qualify why the CPU would cause lag in a game. CPUs can cause lag in games, particularly when the game is poorly optimised and doesn’t make use of all cores available. Is the CPU the bottleneck in this situation? Yes it is, because upgrading to a faster CPU with faster single threaded performance (due to increased cache size, faster FSB speeds/lower core multipliers, more efficient, faster core speed etc) will result in less lag, removing the bottleneck. In a perfect world every game would utilise every core of every CPU, however this isn’t the case. Also, blanket statements about “if you have at least XXX processor you won’t experience CPU lag”. While this may or may not be true for the majority of games, that isn’t always the case. Have you tested every CPU/GPU combination with every modern game? Once again, the focus of the article isn’t as high level as your comment. Interestingly, you didn’t detail how the memory controller (as most modern systems memory controllers are situated in the CPU itself). Also, why would you say technically all of the above CPUs are equal? The difference extends beyond the info listed on the Intel site. You also missed out how the E7300 loses out VT support compared to the other two processors but I’ll let that slide. What I can’t let slide is how the E7300 supports the SSE4.1 instruction set whereas the E3400 doesn’t? That’s a fairly large difference in the context of your argument This is another example of the differences between CPUs beyond the obvious (cores, speed, FSB, multiplier L2 cache). To say an E3400 and E7300 is exactly the same besides L2 cache and FSB speed/clock multiplier is ignorant.
Derrick
ლ(ಠ_ಠ )ლ
- Joined
- Nov 22, 2010
- Messages
- 5,085
- Reaction score
- 5
Post 2
System ram being saturated will NOT result in low framerates. It WILL result in the game getting stuck momentarily from time to time as the system has to perform page-file read/write operations. System ram having insufficient bandwidth WILL result in bandwidth constraints between the processor and the graphics card(s), however, and this can have a detrimental effect on the overall performance of a game; this performance will not always translate directly into a lower framerate so much as it will into stuttering (and stuttering, just like lag, is not the same thing as a low framerate. Michael J Fox doesn't move around super slowly, he jitters around). If you want to see an example of what kind of impact this can have and happen to have access to the computers, go take an old dual-core Athlon system or LGA775 dual-core system with 4b of ram on DDR1 and compare it to the same configuration with DDR2 ram and then to one with DDR3 ram. The LGA775 system can cover all three these bases for your testing purposes.
Your first sentence here, really? Will NOT? If I ran a modern system with 1GB of RAM you think that won’t result in consistently low framerates? Please try for me and see what happens.
What you should see is occasional stuttering in games which need to 'stream' high amounts of data from ram to processor and/or graphics card the way the Unreal Engine 3.0 and 3.5 need to do. Examples of games that use this method are Bioshock1/2, Borderlands, the Unreal Tournament games made on the engines, Gears of War 1/2/3.
Now remove half the ram to bring the system down to 2gb and try to run these games at their maximum graphics (and audio) settings. The raw amount of data that needs to be stored for textures, geometry, shaders, AI information, level designs, etc etc etc should quickly saturate the ram (I know that for Borderlands 2gb is barely sufficient at max settings while a 9600GT 512mb in a system with sufficient ram has absolutely no issue handling the game). Prepare to pull your hair out in frustration as these games get stuck at the most inconvenient times and/or level loading takes half a century to complete.
Stuttering I will give you, it is a common issue when the user is low on RAM. As for your example, do the same test with a very power GPU/CPU combination and see if you only get stuttering, or if there is a mix of stuttering and consistently low framerates.
Claiming that the hard-drive is responsible for sudden and potentially lengthy drops in framerate 'before things return to normal' is to be ignorant as to the way games utilize ram and the paging file. You can practically ignore hard-drive performance for nothing other than load times if the game you're playing was able to cache all of its data for any given level into RAM or has good streaming. Want an example of this? Use an old hard-drive with an average of 40mb/s worth of sustained read in the same high-performance system with enough RAM, then compare it to a system with a 80mb/s hard-drive or even a 250mb/s or greater SSD; chances are that there will be no difference except for load times.
Ignorant is one way of putting it. The other would be offering a low level, easy to understand explanation. Combine an aging IDE hard drive (which are more common that you may think) with a low to moderate ( 2 - 4GB) amount of RAM, then load up a game with large levels(I’ll leave this open to examples). Now, replace just the hard drive with a modern SSD and measure framerates, you’re almost guaranteed to see an increase in framerates. I wonder why you failed to mention seek times of an aging hard drive compared to seek times of an SSD. Why didn’t you mention speed at Queue Depths either?
Yet another post-padding article... =_=
You’ll see I often ask why you neglected other “high level” technical aspects of the CPU, GPU and Hard drive. The reason being, they were not part of your arguments or examples, and they were not part of the point you were trying to make. Did you perhaps leave them out the avoid going into an argument that is too technical or in depth? Or did you think them irrelevant in context? I did in my article.
You must remember, articles are written with a specific audience in mind. If you feel that the information is too obvious or too low level then obviously you are a high powered user and not part of the target audience.
I would invite you to follow all of my tips on how to test for lag on a system that suffers from that specific lag (an example would be a high performance system whose graphics card has been replaced with a low end GPU, then turn down graphics settings to see) and see if they addresses the lag (or low framerate, or stuttering, or whatever you use to define it). You should find in the more extreme cases (where upgrading is adviseable) that the how to test advice works.
System ram being saturated will NOT result in low framerates. It WILL result in the game getting stuck momentarily from time to time as the system has to perform page-file read/write operations. System ram having insufficient bandwidth WILL result in bandwidth constraints between the processor and the graphics card(s), however, and this can have a detrimental effect on the overall performance of a game; this performance will not always translate directly into a lower framerate so much as it will into stuttering (and stuttering, just like lag, is not the same thing as a low framerate. Michael J Fox doesn't move around super slowly, he jitters around). If you want to see an example of what kind of impact this can have and happen to have access to the computers, go take an old dual-core Athlon system or LGA775 dual-core system with 4b of ram on DDR1 and compare it to the same configuration with DDR2 ram and then to one with DDR3 ram. The LGA775 system can cover all three these bases for your testing purposes.
Your first sentence here, really? Will NOT? If I ran a modern system with 1GB of RAM you think that won’t result in consistently low framerates? Please try for me and see what happens.
What you should see is occasional stuttering in games which need to 'stream' high amounts of data from ram to processor and/or graphics card the way the Unreal Engine 3.0 and 3.5 need to do. Examples of games that use this method are Bioshock1/2, Borderlands, the Unreal Tournament games made on the engines, Gears of War 1/2/3.
Now remove half the ram to bring the system down to 2gb and try to run these games at their maximum graphics (and audio) settings. The raw amount of data that needs to be stored for textures, geometry, shaders, AI information, level designs, etc etc etc should quickly saturate the ram (I know that for Borderlands 2gb is barely sufficient at max settings while a 9600GT 512mb in a system with sufficient ram has absolutely no issue handling the game). Prepare to pull your hair out in frustration as these games get stuck at the most inconvenient times and/or level loading takes half a century to complete.
Stuttering I will give you, it is a common issue when the user is low on RAM. As for your example, do the same test with a very power GPU/CPU combination and see if you only get stuttering, or if there is a mix of stuttering and consistently low framerates.
Claiming that the hard-drive is responsible for sudden and potentially lengthy drops in framerate 'before things return to normal' is to be ignorant as to the way games utilize ram and the paging file. You can practically ignore hard-drive performance for nothing other than load times if the game you're playing was able to cache all of its data for any given level into RAM or has good streaming. Want an example of this? Use an old hard-drive with an average of 40mb/s worth of sustained read in the same high-performance system with enough RAM, then compare it to a system with a 80mb/s hard-drive or even a 250mb/s or greater SSD; chances are that there will be no difference except for load times.
Ignorant is one way of putting it. The other would be offering a low level, easy to understand explanation. Combine an aging IDE hard drive (which are more common that you may think) with a low to moderate ( 2 - 4GB) amount of RAM, then load up a game with large levels(I’ll leave this open to examples). Now, replace just the hard drive with a modern SSD and measure framerates, you’re almost guaranteed to see an increase in framerates. I wonder why you failed to mention seek times of an aging hard drive compared to seek times of an SSD. Why didn’t you mention speed at Queue Depths either?
Yet another post-padding article... =_=
You’ll see I often ask why you neglected other “high level” technical aspects of the CPU, GPU and Hard drive. The reason being, they were not part of your arguments or examples, and they were not part of the point you were trying to make. Did you perhaps leave them out the avoid going into an argument that is too technical or in depth? Or did you think them irrelevant in context? I did in my article.
You must remember, articles are written with a specific audience in mind. If you feel that the information is too obvious or too low level then obviously you are a high powered user and not part of the target audience.
I would invite you to follow all of my tips on how to test for lag on a system that suffers from that specific lag (an example would be a high performance system whose graphics card has been replaced with a low end GPU, then turn down graphics settings to see) and see if they addresses the lag (or low framerate, or stuttering, or whatever you use to define it). You should find in the more extreme cases (where upgrading is adviseable) that the how to test advice works.
nanonyous
Expert Member
Okay, I went through the trouble of reading all your bold-bits whereas you didn't read my entire article before deciding to address particular paragraphs. While I didn't go into the most technical aspects of memory bandwidth, I did address the memory bandwidth issue for not only the CPU's cache, but also the system ram (and honestly, it doesn't even *matter* that the memory controller is on-die, no matter how technical you want to get - the bandwidth it's capable of is the bandwidth it's capable of), graphics card RAM and even how the CPU, system ram, graphics card and graphics card ram interact with eachother.
I didn't bother going into pci-e lane width since there is quite simply no need to do so; those lacking knowledge that will saturate available PCI-E controller bandwidth using a dual-card solution without knowing what they are doing have probably blown so much money on their motherboard that they simply can't saturate the bandwidth even if they will it to.
I do not 'misunderstand' the 'how to test graphics cards' category; you misunderstand how idiotic your suggestion is by completely taking for granted the fact that physx and similar settings that can be run by nothing but the processor are often found under the graphics settings, and more often than not aren't accompanied by any information indicating that they are going to be running on the processor, simply that they will "cause a severe framerate decrease", which could have absolutely nothing to do with the graphics card.
The E3300/3400, E5400 and E7300 are all physically exactly the same processor. Virtualization has absolutely nothing to do with games performance unless you are planning on running virtual machines in order to run multiple game clients for whatever purpose. By the time a user has the technical ability to do so they will be intelligent enough to look for themselves to determine whether or not their processor supports hardware VT. By the time they are planning to do so, they will have the technical understanding to realise that their system's resources are going to be heavily split up in order to accommodate both the VM and the extra game client running within it.
SSE 4.1 introduced far less than you'd like to give it credit for. Little enough, in fact, that it on its own has no baring on graphics performance for games.
This article is not the one I'm looking for, though it essentially skims the surface of the implementation that was seen in the E7000 series processors. You are perfectly welcome to take the time to find an article or technical document to prove me wrong on this, and I will be glad to admit defeat on this account if I am shown to be wrong.
Until such a time, yes, the aforementioned processors, with exception to L2 cache size, are essentially the same processors. They do not have built-in memory controllers, and as I already mentioned, it doesn't even matter that their memory controllers are not on-die.
As for the stuttering vs low framerates (and please, it's low framerates, not low lag. Lag is only used by gamers and has *nothing* in actual fact to do with framerate, as a computer is still able to maintain a 1:1 time ratio with a framerate of 1fps or greater if necessary - lag would be where the amount of time that transpires outside of the game is longer than that which transpires in the game, and that issue has not existed in most modern computers [read: anything since the GeForce 6000 series or Radeon X18xx series graphics cards] for some time now) on a system with insufficient ram but a sufficiently powerful processor and graphics card combination; no. You are plain wrong here. I have tested and have tested extensively the impact that ram capacity and ram bandwidth has on performance in games, and the only situation where ram will cause a drop in overall framerate is when the engine is utilizing a lot of stream of content, which as I already explained in some detail in my post way up there, will be reliant on memory bandwidth, NOT capacity. If the content that needs to be streamed is not in ram at that time due to needing to be deferred for other information, then that information needs to be drawn from the hard-drive again, which is where the 'sudden drop in framerate before things return to normal' comes in. More realistically, the game will simply 'stop' while it's loading those resources from the hard-drive and going berserk with the page file.
Which brings us to the hard-drive issue. No, the hard-drive will NOT cause low framerates. Ever. It will increase 'stuttering', load times and the duration of those 'sudden drops in framerate', but this will only carry on for as long as the computer has to continue reading data from the drive and performing read/write operations on the page file.
Your suggestion of replacing the hard-drive with an SSD is, quite frankly, !@#$ing stupid. You're telling someone to go out and spend insane amounts of money on a hard-drive that will give them some kind of magical improvement in performance when the performance issue they're encountering is far more likely to be the cause of a memory capacity constraint and/or bandwidth issue between the cpu/ram/graphics card. I don't know what games you're playing, but most modern games these days are able to cache virtually all of a level's data into ram (and data for several levels ahead seeing as a lot of the data, such as models, sounds, shaders, textures etc etc etc are being reused between levels) assuming there is enough ram capacity for it all. A 40mb/s average read speed drive is going to do nothing to your framerate vs using a 500mb/s SSD in the same system other than make your wallet feel smaller, your epeen feel bigger and your ego fall somewhere in between.
It's far more sensible to advise that people look into replacing (not increasing the capacity of, since you're so keen on keeping things super simple; we'll keep things super simple to reduce complications [that is to say, we won't take into account the potential issues a memory controller could run into with trying to use differing capacities of ram, double/single-sided ram sticks, slot configuration restrictions, memory age and thus potential damage it might have suffered, etc etc etc]) their ram with a set of higher capacity sticks so that they shouldn't run out of ram unless they're doing something very wrong in the background. It'll certainly be a far less expensive upgrade for them to perform than going out to buy a SSD, and it will give them a real-world tangible performance increase for all things they could use their computer for vs just the SSD.
As for seek times; really? What do you think seek times on a hard-drive have to do with framerates in a game? Quicker seek times allow for quicker read/write operations of extremely small chunks as opposed to sequential read/write operations. Neither of these, again referring to the multiple pieces about memory and its impact on game performance, will magically give you an increase in framerate unless you're doing it wrong with your ram.
As for shader processors/CUDA cores, ROPs etc - one can practically ignore these on the basis of only needing to know roughly what the horsepower of the graphical processing component of the graphics card is in relation to other graphics cards it's being stacked up against. Beyond that, it's only important to understand what impact graphics card memory capacities and memory bandwidth can have on your overall performance. You don't need to to go into the most nitty-gritty details of what each component's function is and how it stacks up against a previous/newer generation's iteration or replacement architecture is or what the competing architecture is, how it functions and how it stacks up. All you need to understand is the basic impact that generation X's Y configuration with Z memory configuration should have on your gaming performance when taking graphics settings and resolution into account for any specific title.
So to summarize, since I was essentially addressing you and your article:
1. Yes, articles have a target audience
2. No, that is NOT an excuse to not go into detail
3. No, that is NOT an excuse to give wholly misleading information
4. No, that is NOT an excuse to assume that you can just give a crappy (yes, crappy blanket coverage of all the bases and expect it to be at all relevant at the end of the day
5. If you're going to be writing up more detailed articles concerning your points then you need to indicate that in your article, either at the beginning or the end as a "this is a preview of topics that are going to be covered in upcoming articles regarding computer hardware upgrades for games yada yada" or "this was just a sneak peak at upcoming articles to better guide you on identifying bottlenecks and determining an upgrade path yada yada". Your article did neither. It presents itself as a self-contained, all-inclusive, nothing-more-to-see-here-folks article.
I could compile a list of links to Tomshardware or Overclock.net articles to pick two sites off the top of my head that would give far more useful information pertaining to exactly these topics that are still explained in a simple enough manner that the lay man would be able to understand them and more importantly, would probably walk away feeling the richer for having taught themselves something new about computer hardware and how it affects their usage of their computer (and more importantly, how not to blow stupid amounts of money on something that won't necessarily truly benefit them, like an SSD). It is because of exactly this fact that I find fault with your article. It is because I feel that I have far more experience and knowledge than you regarding this subject that I decided to tackle it, without going into TOO much detail.
I didn't bother going into pci-e lane width since there is quite simply no need to do so; those lacking knowledge that will saturate available PCI-E controller bandwidth using a dual-card solution without knowing what they are doing have probably blown so much money on their motherboard that they simply can't saturate the bandwidth even if they will it to.
I do not 'misunderstand' the 'how to test graphics cards' category; you misunderstand how idiotic your suggestion is by completely taking for granted the fact that physx and similar settings that can be run by nothing but the processor are often found under the graphics settings, and more often than not aren't accompanied by any information indicating that they are going to be running on the processor, simply that they will "cause a severe framerate decrease", which could have absolutely nothing to do with the graphics card.
The E3300/3400, E5400 and E7300 are all physically exactly the same processor. Virtualization has absolutely nothing to do with games performance unless you are planning on running virtual machines in order to run multiple game clients for whatever purpose. By the time a user has the technical ability to do so they will be intelligent enough to look for themselves to determine whether or not their processor supports hardware VT. By the time they are planning to do so, they will have the technical understanding to realise that their system's resources are going to be heavily split up in order to accommodate both the VM and the extra game client running within it.
SSE 4.1 introduced far less than you'd like to give it credit for. Little enough, in fact, that it on its own has no baring on graphics performance for games.
This article is not the one I'm looking for, though it essentially skims the surface of the implementation that was seen in the E7000 series processors. You are perfectly welcome to take the time to find an article or technical document to prove me wrong on this, and I will be glad to admit defeat on this account if I am shown to be wrong.
Until such a time, yes, the aforementioned processors, with exception to L2 cache size, are essentially the same processors. They do not have built-in memory controllers, and as I already mentioned, it doesn't even matter that their memory controllers are not on-die.
As for the stuttering vs low framerates (and please, it's low framerates, not low lag. Lag is only used by gamers and has *nothing* in actual fact to do with framerate, as a computer is still able to maintain a 1:1 time ratio with a framerate of 1fps or greater if necessary - lag would be where the amount of time that transpires outside of the game is longer than that which transpires in the game, and that issue has not existed in most modern computers [read: anything since the GeForce 6000 series or Radeon X18xx series graphics cards] for some time now) on a system with insufficient ram but a sufficiently powerful processor and graphics card combination; no. You are plain wrong here. I have tested and have tested extensively the impact that ram capacity and ram bandwidth has on performance in games, and the only situation where ram will cause a drop in overall framerate is when the engine is utilizing a lot of stream of content, which as I already explained in some detail in my post way up there, will be reliant on memory bandwidth, NOT capacity. If the content that needs to be streamed is not in ram at that time due to needing to be deferred for other information, then that information needs to be drawn from the hard-drive again, which is where the 'sudden drop in framerate before things return to normal' comes in. More realistically, the game will simply 'stop' while it's loading those resources from the hard-drive and going berserk with the page file.
Which brings us to the hard-drive issue. No, the hard-drive will NOT cause low framerates. Ever. It will increase 'stuttering', load times and the duration of those 'sudden drops in framerate', but this will only carry on for as long as the computer has to continue reading data from the drive and performing read/write operations on the page file.
Your suggestion of replacing the hard-drive with an SSD is, quite frankly, !@#$ing stupid. You're telling someone to go out and spend insane amounts of money on a hard-drive that will give them some kind of magical improvement in performance when the performance issue they're encountering is far more likely to be the cause of a memory capacity constraint and/or bandwidth issue between the cpu/ram/graphics card. I don't know what games you're playing, but most modern games these days are able to cache virtually all of a level's data into ram (and data for several levels ahead seeing as a lot of the data, such as models, sounds, shaders, textures etc etc etc are being reused between levels) assuming there is enough ram capacity for it all. A 40mb/s average read speed drive is going to do nothing to your framerate vs using a 500mb/s SSD in the same system other than make your wallet feel smaller, your epeen feel bigger and your ego fall somewhere in between.
It's far more sensible to advise that people look into replacing (not increasing the capacity of, since you're so keen on keeping things super simple; we'll keep things super simple to reduce complications [that is to say, we won't take into account the potential issues a memory controller could run into with trying to use differing capacities of ram, double/single-sided ram sticks, slot configuration restrictions, memory age and thus potential damage it might have suffered, etc etc etc]) their ram with a set of higher capacity sticks so that they shouldn't run out of ram unless they're doing something very wrong in the background. It'll certainly be a far less expensive upgrade for them to perform than going out to buy a SSD, and it will give them a real-world tangible performance increase for all things they could use their computer for vs just the SSD.
As for seek times; really? What do you think seek times on a hard-drive have to do with framerates in a game? Quicker seek times allow for quicker read/write operations of extremely small chunks as opposed to sequential read/write operations. Neither of these, again referring to the multiple pieces about memory and its impact on game performance, will magically give you an increase in framerate unless you're doing it wrong with your ram.
As for shader processors/CUDA cores, ROPs etc - one can practically ignore these on the basis of only needing to know roughly what the horsepower of the graphical processing component of the graphics card is in relation to other graphics cards it's being stacked up against. Beyond that, it's only important to understand what impact graphics card memory capacities and memory bandwidth can have on your overall performance. You don't need to to go into the most nitty-gritty details of what each component's function is and how it stacks up against a previous/newer generation's iteration or replacement architecture is or what the competing architecture is, how it functions and how it stacks up. All you need to understand is the basic impact that generation X's Y configuration with Z memory configuration should have on your gaming performance when taking graphics settings and resolution into account for any specific title.
So to summarize, since I was essentially addressing you and your article:
1. Yes, articles have a target audience
2. No, that is NOT an excuse to not go into detail
3. No, that is NOT an excuse to give wholly misleading information
4. No, that is NOT an excuse to assume that you can just give a crappy (yes, crappy blanket coverage of all the bases and expect it to be at all relevant at the end of the day
5. If you're going to be writing up more detailed articles concerning your points then you need to indicate that in your article, either at the beginning or the end as a "this is a preview of topics that are going to be covered in upcoming articles regarding computer hardware upgrades for games yada yada" or "this was just a sneak peak at upcoming articles to better guide you on identifying bottlenecks and determining an upgrade path yada yada". Your article did neither. It presents itself as a self-contained, all-inclusive, nothing-more-to-see-here-folks article.
I could compile a list of links to Tomshardware or Overclock.net articles to pick two sites off the top of my head that would give far more useful information pertaining to exactly these topics that are still explained in a simple enough manner that the lay man would be able to understand them and more importantly, would probably walk away feeling the richer for having taught themselves something new about computer hardware and how it affects their usage of their computer (and more importantly, how not to blow stupid amounts of money on something that won't necessarily truly benefit them, like an SSD). It is because of exactly this fact that I find fault with your article. It is because I feel that I have far more experience and knowledge than you regarding this subject that I decided to tackle it, without going into TOO much detail.
nanonyous
Expert Member
You ask if I've used xyz hardware combinations before. Yes, I have. I have three E3300/E3400 systems here right now along with my i3 based system. I have spent some quality time with a Phenom X4. I have used every memory standard from freaking Vic20 64kb memory cartridges (here's looking at you, Jab) to computers that used some kind of memory consisting of loose chips you stuck into sockets on the motherboard to PC33 SIMMS to DIMMS to DDR1/2/3. I have used systems with integrated IBM graphics chips up to computers with 1mb monochrome display cards (no, not graphics cards, display cards, they did nothing but render the information into a format that could be displayed on a CRT) to the first 2mb graphics accelerators, then a Voodoo 4mb, a Voodoo Rush 6mb, a Vooodoo Banshee 16mb, a TNT Vanta, TNT2 Ultra, TNT2 Ultra64, GeForce2 64mb MX400, Radeon 9200SE 128mb, various other Radeon9200 incarnations, GeForce 3s, Geforce 4s, GeForce 5s (though I wish I hadn't), Radeon 9500 in plain, SE and Pro variations, Radeon 9700 in plain, Pro and AIW variations (I miss my AIW), GeForce 6600 and 6800 in various incarnations, various IGPs from SiS/Virge, VIA and Intel, GeForce 7300, GeForce 7600GT and 7800GT, GeForce 8600GT, 8800GTS 640mb, 8800GT, 9600GT, GTX460 1024mb (current card) and GT430, and briefly got do test out a X1800 and HD4870.
I've personally used and tested all manner of processors coming up that same timeline with all manner of hard-drives in different RAID configurations and with different drive performance levels, along with a few high-RPM drives here and there.
I feel I have used computers for long enough, being that I've been gaming for the past 21 years on them, and been keeping myself up-to-date enough on the information that actually matters to tell that based on your response to me, you are in ways probably in over your head. I feel I have used enough different hardware configurations, built, mixed, broken and fiddled with enough different configurations/components to be able to ascertain within a few minutes where the performance issues on a computer may be.
Do you know what the first thing is I do when I hop onto any computer I've never used before? I get to the processor and memory information to figure out what kind of processing power I can expect and what memory capacity I'm working with. Afterwards, if it's at all relevant to what I plan on doing on the computer, I may check its graphics card. Motherboard? It hardly matters. If I'm going to be running games on it I'll probably be familiar enough with the game in question and the components in use that I should be able to understand which settings will have what impact on the game's performance.
It's because of all of this that I feel you wrote up nothing but a post-padding article that serves no purpose. You could just as well have left it at the very first line in the article, namely:
And people could have walked away as enlightened as though they had read the entire thing.
I may not be employed with a company nor do I have credentials to the effect of, but I consider myself to be and employ myself on a freelance basis as a computer hardware consultant. If new technology is coming out and someone needs me to build a system for them, I make a point of finding out about what is currently available and what may be available in the near future to ascertain how any given configuration may suit my client in the next year or two. I've more recently begun making a point of figuring out what price/performance ratios on a per-component basis are sensible assuming my client is willing to spend X amount on a system, explaining to them in no limited detail that if they get themselves Y system they will be able to get themselves Z upgrade next year and the year thereafter, at which point they will have spent the same as X, but will have in all likelihood come out with more than 2x the ultimate performance of their third configuration.
I'm going to leave it at this for now so that I can get some sleep. I hope I've been verbose enough to cover all required bases at this point.
I've personally used and tested all manner of processors coming up that same timeline with all manner of hard-drives in different RAID configurations and with different drive performance levels, along with a few high-RPM drives here and there.
I feel I have used computers for long enough, being that I've been gaming for the past 21 years on them, and been keeping myself up-to-date enough on the information that actually matters to tell that based on your response to me, you are in ways probably in over your head. I feel I have used enough different hardware configurations, built, mixed, broken and fiddled with enough different configurations/components to be able to ascertain within a few minutes where the performance issues on a computer may be.
Do you know what the first thing is I do when I hop onto any computer I've never used before? I get to the processor and memory information to figure out what kind of processing power I can expect and what memory capacity I'm working with. Afterwards, if it's at all relevant to what I plan on doing on the computer, I may check its graphics card. Motherboard? It hardly matters. If I'm going to be running games on it I'll probably be familiar enough with the game in question and the components in use that I should be able to understand which settings will have what impact on the game's performance.
It's because of all of this that I feel you wrote up nothing but a post-padding article that serves no purpose. You could just as well have left it at the very first line in the article, namely:
And people could have walked away as enlightened as though they had read the entire thing.
I may not be employed with a company nor do I have credentials to the effect of, but I consider myself to be and employ myself on a freelance basis as a computer hardware consultant. If new technology is coming out and someone needs me to build a system for them, I make a point of finding out about what is currently available and what may be available in the near future to ascertain how any given configuration may suit my client in the next year or two. I've more recently begun making a point of figuring out what price/performance ratios on a per-component basis are sensible assuming my client is willing to spend X amount on a system, explaining to them in no limited detail that if they get themselves Y system they will be able to get themselves Z upgrade next year and the year thereafter, at which point they will have spent the same as X, but will have in all likelihood come out with more than 2x the ultimate performance of their third configuration.
I'm going to leave it at this for now so that I can get some sleep. I hope I've been verbose enough to cover all required bases at this point.
Derrick
ლ(ಠ_ಠ )ლ
- Joined
- Nov 22, 2010
- Messages
- 5,085
- Reaction score
- 5
I'll be honest, I just read through all of that as daunting as it was.
As for knowing more about hardware/technology, feel feel to think that, there is actually a good possibility that it's true. There is always someone who knows more about tech that me or you or any old Joe soap. As for experience, I could go on about how that hardware list you mentioned doesn't even cover a tenth of the hardware I have experience with etc etc, but there is very little point to getting into a pissing match about whose credentials are more impressive, this is the internet after all.
So my reply(short though it may be, I need to get some work done) to you is, I really do appreciate your comments on my articles and look forward to reading them in future.
As long as you're willing to comment I'm willing to read them and take them as constructive criticism (even though your abrasive tone to me not to mention other members in similar thread sounds like anything but constructive).
As for knowing more about hardware/technology, feel feel to think that, there is actually a good possibility that it's true. There is always someone who knows more about tech that me or you or any old Joe soap. As for experience, I could go on about how that hardware list you mentioned doesn't even cover a tenth of the hardware I have experience with etc etc, but there is very little point to getting into a pissing match about whose credentials are more impressive, this is the internet after all.
So my reply(short though it may be, I need to get some work done) to you is, I really do appreciate your comments on my articles and look forward to reading them in future.
As long as you're willing to comment I'm willing to read them and take them as constructive criticism (even though your abrasive tone to me not to mention other members in similar thread sounds like anything but constructive).nanonyous
Expert Member
I'm abrasive because I hate it when people expect I should try and treat them as softly as possible while explaining why they are wrong. Often it has happened that when I've tried to do that, people simply end up assuming I'm treating them as the town simpleton, even if that wasn't my intention; so why bother sugar-coating to get a point across?
I've said it before though, I do appreciate that MyBroadband and its columnists/editors are trying to create a tech news oriented site to compete with other international sites, but I don't like that there are potentially misleading, useless or otherwise wasteful posts.
The three posts of yours that I found particularly annoying in this regard, more recently, were the ones regarding mousepads, '5 gimmicks you'd like to own' and this article.
If you'd like to have summaries of information, then by all means. Simply ensure that there is more detailed information to go with it for those that may be reading to who it's at all relevant. Telling people "Your system is laggy? Better go out and spend money on hardware. But wait! You can do all these tests that won't really give you an idea as to what part is slow! Better just upgrade ALL the parts!"
It's the equivalent of saying that if your car is starting to perform a bit slowly, you can perform a few tests such as sitting on a hill in first gear and seeing how well your car is able to get up the hill if you simply floor it; are your wheels spinning? Better get new tires. Is the car struggling to get up there? Might be the fuel/air mixture. It could be the differential... or it could be that the timing is out... (these are things I know very little to nothing about, but what I do know is that if the wheels are simply spinning [assuming the torque on the wheels is relatively tame] that that means your tires don't have sufficient traction on the surface, which could either mean that your tires are bad, that you simply don't have enough of a contact patch [which could be a factor of tire width, wheel and tire size, tire stiffness, tread and the surface on which the car is resting, not to mention the surface consistency to begin with]). To the average layman, that test would be useless.
Now on the other hand, if you explained to them how their tires alone could have an effect on the aforementioned test, they can better isolate the tires to ensure that they aren't responsible for the car's failure to get up the hill in the expected manner.
And I still honestly find your claims regarding the hard-drive to be quite ridiculous; especially that you recommend that people go out and get an SSD as though it's some silver bullet.
I'm going to guess you've read the same article as well, but here's the link for those that haven't that might be reading this thread.
It's very important to note that they are using a wholly average, otherwise known as 'mainstream' system for their test, not an enthusiast rig with multiple dual-gpu cards, the greatest processor on earth and a motherboard to go with it (not to mention enough ram to make you dizzy).
The gist of it boils down to the hard-drive, during gameplay, not necessarily having that great an impact on overall gameplay performance. Yes, you may get stuttering. No, you are not going to get 'lag' (a reduction in overall framerate).
The most important factor to take into account though, as I've already pointed out, is the fact that a SSD costs that much more than increasing the amount of ram your system has to (at least) reduce the amount of stuttering you'll experience during gameplay from data reads (not writes) that may not already be in ram; this ignoring the fact that the primary benefit of an SSD simply doesn't correlate to reads for game data, seeing as the bulk of it for virtually all games comes down to large amounts of sequentially read data.
Some of the rest of that amount of money saved could be put towards other immediate or future upgrades, and it's because of exactly that that I found this latest article of yours annoying; again the fact that what you essentially end up saying is "slow pc? go out and blow money on uninformed purchases."
I look forward to these articles you say you're going to write up that will go into more detail so that I may be as abrasive of tone when commenting on them as well (if the need arises).
ps. sorry if my posts are long-winded; I type very quickly and prefer to cover all bases so that there's less room for misinterpretation per post
I've said it before though, I do appreciate that MyBroadband and its columnists/editors are trying to create a tech news oriented site to compete with other international sites, but I don't like that there are potentially misleading, useless or otherwise wasteful posts.
The three posts of yours that I found particularly annoying in this regard, more recently, were the ones regarding mousepads, '5 gimmicks you'd like to own' and this article.
If you'd like to have summaries of information, then by all means. Simply ensure that there is more detailed information to go with it for those that may be reading to who it's at all relevant. Telling people "Your system is laggy? Better go out and spend money on hardware. But wait! You can do all these tests that won't really give you an idea as to what part is slow! Better just upgrade ALL the parts!"
It's the equivalent of saying that if your car is starting to perform a bit slowly, you can perform a few tests such as sitting on a hill in first gear and seeing how well your car is able to get up the hill if you simply floor it; are your wheels spinning? Better get new tires. Is the car struggling to get up there? Might be the fuel/air mixture. It could be the differential... or it could be that the timing is out... (these are things I know very little to nothing about, but what I do know is that if the wheels are simply spinning [assuming the torque on the wheels is relatively tame] that that means your tires don't have sufficient traction on the surface, which could either mean that your tires are bad, that you simply don't have enough of a contact patch [which could be a factor of tire width, wheel and tire size, tire stiffness, tread and the surface on which the car is resting, not to mention the surface consistency to begin with]). To the average layman, that test would be useless.
Now on the other hand, if you explained to them how their tires alone could have an effect on the aforementioned test, they can better isolate the tires to ensure that they aren't responsible for the car's failure to get up the hill in the expected manner.
And I still honestly find your claims regarding the hard-drive to be quite ridiculous; especially that you recommend that people go out and get an SSD as though it's some silver bullet.
Tom's Hardware said:
I'm going to guess you've read the same article as well, but here's the link for those that haven't that might be reading this thread.
It's very important to note that they are using a wholly average, otherwise known as 'mainstream' system for their test, not an enthusiast rig with multiple dual-gpu cards, the greatest processor on earth and a motherboard to go with it (not to mention enough ram to make you dizzy).
The gist of it boils down to the hard-drive, during gameplay, not necessarily having that great an impact on overall gameplay performance. Yes, you may get stuttering. No, you are not going to get 'lag' (a reduction in overall framerate).
The most important factor to take into account though, as I've already pointed out, is the fact that a SSD costs that much more than increasing the amount of ram your system has to (at least) reduce the amount of stuttering you'll experience during gameplay from data reads (not writes) that may not already be in ram; this ignoring the fact that the primary benefit of an SSD simply doesn't correlate to reads for game data, seeing as the bulk of it for virtually all games comes down to large amounts of sequentially read data.
Some of the rest of that amount of money saved could be put towards other immediate or future upgrades, and it's because of exactly that that I found this latest article of yours annoying; again the fact that what you essentially end up saying is "slow pc? go out and blow money on uninformed purchases."
I look forward to these articles you say you're going to write up that will go into more detail so that I may be as abrasive of tone when commenting on them as well (if the need arises).
ps. sorry if my posts are long-winded; I type very quickly and prefer to cover all bases so that there's less room for misinterpretation per post
Last edited:
ingeon
Senior Member
- Joined
- Apr 11, 2008
- Messages
- 752
- Reaction score
- 4
Go for it
Powersupply calculator estimator
Very simply you have 3 basic voltage rails. 3V, 5V, 12V. Each piece of hardware (CPU/RAM/GPU/HDD) needs
a certain power to operate optimally.
If the supply is to small it will choke (thus goes into current limit or power limit depending on whether it`s
a constant current or constant power source) One would not necessarily experience problems until
running software/game that fully utilize the hardware limits. HWiNFO is a great way to check this. Start it up, run the game that makes your PC stutter, exit the game and compare the "current" and minimum voltage values. There should hardly be a difference. (Just make sure HWiNFO supports your chipsets)
On a side note, from a temperature view. For some strange reason the newer hardware seems to be more fussy.
This weekend we had two gaming PC`s that randomly stuttered. We installed HWiNFO and while playing the CPU that normally sits on 40C idle, while gaming went well past 80C. I fetched my blower, cleaned the whole machine (blocking the fans) and the max temp dropped to 63C. Problem sorted. (His PC was really really dirty)
Last edited:
nanonyous
Expert Member
Intel CPUs (I'm sure AMD has an equivalent, I just haven't cared enough to notice of late since they seldom run too hot in my experience) have thermal throttling that kicks in at certain temperatures for certain processors, with higher and lower thresholds based on specific model.
I believe the Core2Duo processors, as an example, and just to generalize, would throttle from 83c. By the time the processor hit 100c it would throttle more severely, while 105c would be the emergency shutdown temperature (the numbers are probably off, cba to check)
A dodgy powersupply on its own won't cause stuttering, though. If your processor, RAM, motherboard and its chipset(s) receive insufficient power, the system will simply freeze, bsod (more often spikes will cause this as opposed to dips) or restart.
A graphics card with external power, however, is a completely different matter. Most graphics cards can function without any external power whatsoever, they simply won't be able to operate at full potential simply because the auxiliary power required isn't available to feed all the hardware on the card.
I'm not familiar with the tool you linked, but Speedfan can also monitor voltages on almost all motherboards made after 2004 with voltage monitoring in their chipsets. I don't think I've encountered a single budget-board these days that doesn't have monitoring for the most important bits, either.
I believe the Core2Duo processors, as an example, and just to generalize, would throttle from 83c. By the time the processor hit 100c it would throttle more severely, while 105c would be the emergency shutdown temperature (the numbers are probably off, cba to check)
A dodgy powersupply on its own won't cause stuttering, though. If your processor, RAM, motherboard and its chipset(s) receive insufficient power, the system will simply freeze, bsod (more often spikes will cause this as opposed to dips) or restart.
A graphics card with external power, however, is a completely different matter. Most graphics cards can function without any external power whatsoever, they simply won't be able to operate at full potential simply because the auxiliary power required isn't available to feed all the hardware on the card.
I'm not familiar with the tool you linked, but Speedfan can also monitor voltages on almost all motherboards made after 2004 with voltage monitoring in their chipsets. I don't think I've encountered a single budget-board these days that doesn't have monitoring for the most important bits, either.
Kosmik
Honorary Master
Holy ****.
Lol at the pics, I see a Banshee card and I'd suspect that to be a MFM hardrive, maybe a 10Meg?
Lol at the pics, I see a Banshee card and I'd suspect that to be a MFM hardrive, maybe a 10Meg?