The PC Build Thread

Status
Not open for further replies.
Progenix Oj101 said:
I have an issue with this, and the way every benchmark is run.

Who plays games on a fresh install of Windows? Be honest, every time you play a new game do you reload Windows and NOTHING else other than required drivers? None of the following?

Steam
Uplay
Epic Games Launcher
Team Viewer
Skype
Teams
Zoom
YouTube
iCUE
MSI Afterburner
Plex
A Linux distro downloader of choice
Streaming software
etc etc etc?

If you have any of the above (or anything else installed), do you also close EVERYTHING before launching a game?

I don't know anyone who uses a computer that way, and therefore I don't find videos like the above or even most (all) benchmarks to be of much relevance unless all you want is a side-by-side performance comparison in a best case scenario.

I'm sitting at 13.1GB RAM usage right now - mostly Chrome and Photoshop, along with Skype/iCUE/Sound Blaster control panel/Afterburner/Outlook/printer utilities/Hamachi/Plex/Google Drive running in the background. I'm not closing anything before firing up a game. How happy do you think most games will be with less than 3GB free memory?
Click to expand...
Can't imagine firing up a game with outlook and Photoshop running. If you're that reckless with resources then sure, you'd need more.

But that's also just silly, so not sure it applies.

I don't really close anything if I'm gaming, but I wouldn't do it with Office, Powerpoint or other resource-hungry app running.
 
I have Steam, GeForce Experience, GCC, Chrome with 4 tabs and a Youtube video playing at 1440p and I'm sitting at 6.3GB usage.
 
UrBaN963 said:
Can't imagine firing up a game with outlook and Photoshop running. If you're that reckless with resources then sure, you'd need more.

But that's also just silly, so not sure it applies.

I don't really close anything if I'm gaming, but I wouldn't do it with Office, Powerpoint or other resource-hungry app running.
Click to expand...
My list of things which DO NOT get closed for any reason:

Epic Games Launcher: 80MB
Google Drive: 100MB
Hamachi: 10MB
iCUE: 280MB
NVIDIA utilities: 20MB
Plex: 2GB (I assume it's indexing - although that's something that can happen at any time so...)
Printer utilities: 200MB
Skype: 350MB
Sound Blaster utility: 25MB
Steam: 725MB (most for an instance of steamwebhelper.exe)
Sticky Notes: 10MB
Team Viewer: 40MB

That's already 4GB over and above Windows itself. Add around 4-4.5GB for that, and your 16GB is down to less than 8GB usable.

Add to that Photoshop (I use it more than the built-in image viewer so its permanently open - actually in my startup items), Outlook, and Chrome and there's nothing left for games to breathe with 16GB RAM.

RAM is cheap enough. Why inconvenience yourself by not having enough, unless budget really doesn't allow?
 
Progenix Oj101 said:
My list of things which DO NOT get closed for any reason:

Epic Games Launcher: 80MB
Google Drive: 100MB
Hamachi: 10MB
iCUE: 280MB
NVIDIA utilities: 20MB
Plex: 2GB (I assume it's indexing - although that's something that can happen at any time so...)
Printer utilities: 200MB
Skype: 350MB
Sound Blaster utility: 25MB
Steam: 725MB (most for an instance of steamwebhelper.exe)
Sticky Notes: 10MB
Team Viewer: 40MB

That's already 4GB over and above Windows itself. Add around 4-4.5GB for that, and your 16GB is down to less than 8GB usable.

Add to that Photoshop (I use it more than the built-in image viewer so its permanently open - actually in my startup items), Outlook, and Chrome and there's nothing left for games to breathe with 16GB RAM.

RAM is cheap enough. Why inconvenience yourself by not having enough, unless budget really doesn't allow?
Click to expand...
You perhaps missed the initial post and why I suggested it is not necessary.

My argument wasn't that it was never necessary, my argument was for that person, with that rig, 16GB wasn't going to make much difference, if any, to his performance.

Get 32GB if you can, but let's not pretend it's necessary. It isn't necessary.
 
UrBaN963 said:
You perhaps missed the initial post and why I suggested it is not necessary.

My argument wasn't that it was never necessary, my argument was for that person, with that rig, 16GB wasn't going to make much difference, if any, to his performance.

Get 32GB if you can, but let's not pretend it's necessary. It isn't necessary.
Click to expand...
My post was aimed at the video and why it's pointless, not at any one person in particular.
 
Progenix Oj101 said:
My post was aimed at the video and why it's pointless, not at any one person in particular.
Click to expand...
Ah, so we've just gone off on a tangent basically.

The video is valid as it removes as many variables as possible. Still, as you say, nobody runs with nothing open, though nobody should be gaming with what you have open.
 
UrBaN963 said:
Ah, so we've just gone off on a tangent basically.

The video is valid as it removes as many variables as possible. Still, as you say, nobody runs with nothing open, though nobody should be gaming with what you have open.
Click to expand...
With what I have open isn't even on the extreme side. The average is likely somewhere in the middle, as the second you have just ONE thing open in the background it makes the video invalid by anything from a small to massive margin.

/shrug
 
Progenix Oj101 said:
With what I have open isn't even on the extreme side. The average is likely somewhere in the middle, as the second you have just ONE thing open in the background it makes the video invalid by anything from a small to massive margin.

/shrug
Click to expand...
No, it doesn't. It provides a baseline from which to work, which would have to be adapted to your use case and personal circumstances.

It's a fairly close reflection of my usage, though obviously quite far from yours.
 
UrBaN963 said:
No, it doesn't. It provides a baseline from which to work, which would have to be adapted to your use case and personal circumstances.
Click to expand...
To an extent, maybe. The difference between a fresh install with 16GB RAM leaving 12GB usable for the game could be day and night vs an older (more typical*) install with 16GB RAM leaving 9GB usable for the game.

It COULD have been an amazing test if they used maxmem to increase the available RAM in 1GB/2GB increments and viewers could compare their actual available RAM to that part of the video. This cannot be done with any measure of accuracy when the test jumps in increments of 8/16GB.

UrBaN963 said:
It's a fairly close reflection of my usage, though obviously quite far from yours.
Click to expand...
There's no such thing as typical though, it just happens to align with yours. There are already several people here saying their PC, while idle, is using 10+GB RAM.

People who JUST want to game can get a console or even Steam Deck for a satisfying experience. The majority of PC users do far more than just game, and will have other apps and tasks running in the background. Regardless of what those tasks are, they make a comparison to the video impossible. Do you compare your 12GB free to the 16GB graph or the 8GB graph? Maybe 13GB free will be 1:1 with the 16GB graph, and due to an increase in swapping having 12.5GB free will put you 1:1 with the 8GB graph. The scaling might not be linear at all.

I think 1GB or even 2GB granularity would have added a ton of value to the tests.
 
Progenix Oj101 said:
To an extent, maybe. The difference between a fresh install with 16GB RAM leaving 12GB usable for the game could be day and night vs an older (more typical*) install with 16GB RAM leaving 9GB usable for the game.

It COULD have been an amazing test if they used maxmem to increase the available RAM in 1GB/2GB increments and viewers could compare their actual available RAM to that part of the video. This cannot be done with any measure of accuracy when the test jumps in increments of 8/16GB.


There's no such thing as typical though, it just happens to align with yours. There are already several people here saying their PC, while idle, is using 10+GB RAM.

People who JUST want to game can get a console or even Steam Deck for a satisfying experience. The majority of PC users do far more than just game, and will have other apps and tasks running in the background. Regardless of what those tasks are, they make a comparison to the video impossible. Do you compare your 12GB free to the 16GB graph or the 8GB graph? Maybe 13GB free will be 1:1 with the 16GB graph, and due to an increase in swapping having 12.5GB free will put you 1:1 with the 8GB graph. The scaling might not be linear at all.

I think 1GB or even 2GB granularity would have added a ton of value to the tests.
Click to expand...
Sure, there will always be ways to improve the testing and there will always be differing scenarios. Tests like that have value for what they are, but they cannot ever be taken as gospel because there will always be discrepancies. Some degree of intelligence has to be applied to their consumption.
 
UrBaN963 said:
Some degree of intelligence has to be applied to their consumption.
Click to expand...
Intelligence doesn't cover non-linear scaling. At what exact point (and there will be one) does swapping to storage become and issue? We don't need to know it to the bit level, but at least 1-2GB would be FAR more useful than two tests 16 GIGABYTES apart.

If we take the BF5 test on an RX 570, for all you know just loading Discord in the background is enough to drop from a minimum of 61 FPS all the way down to 54. With that level of hardware, it's unlikely that you'll be using a G-Sync/FreeSync/Adaptive Sync monitor, and with V-Sync enabled on a 60Hz monitor that'll drop you down to 30 FPS.

Do you just need to close Discord? Do you need to close Chrome, Discord, NVIDIA control center, iCUE, Plex, uTorrent, Skype, Teams and and Steam? Maybe you Alt/Tab out of the game while waiting to respawn/camping and check on Discord/Skype/Teams/WhatsApp/whatever. Maybe you have music playing in the background on YouTube. Maybe you're downloading a series to watch when your game ends. Maybe you don't close Steam because often it takes inexplicably long to launch. Yes, these are all "maybes" but they're completely realistic maybes.

The level of granularity (or lack thereof) given means that 100% of the testing falls on the user, rather than the user being given a reasonable target to aim for.

"16GB is enough according to that test - I have 16GB with these three apps open that I use in the background (maybe required for DPI adjustments on your mouse because it isn't supported on a hardware level on your mouse, maybe Afterburner for your custom fan curve, etc) but I'm getting 8GB performance - WHY?!!" vs "I see that once free RAM drops from 12GB to 10GB, performance falls through the roof - I need to just put up with a bit of extra fan noise and stick to one DPI"

The above is exaggerated for effect, but no "normal" user expects to check the 8GB graph when they're running 16GB because the performance drop happens very close to having less than 16GB available.

It's not like a graphics card where if your core speed is 5% lower you can expect a linear (even though it is seldom the case, it's generally quite close) drop in performance. You can't look at that chart and say "the pattern is 54 to 61 in 8x1GB steps, so with 10GB free I will get 56 FPS." Hell, we don't even have an available RAM (AVAILABLE, not INSTALLED) baseline to begin with. What could very well be happening is more like 6GB free? 54 FPS. 10GB free? 54 FPS. 12GB free? 54 FPS. 13GB free? 61 FPS.

I specifically cherry picked that result from the entire video because it shows the greatest flaw in testing where the drop is 60 to 30 FPS with V-Sync enabled.

That video COULD have been a winner with not too much additional effort.
 
Gtx Gaming said:
Nope unlocked, I had 90fps before i moved raytracing to max settings.

This is RAW no DLSS or FSR.
Click to expand...
"In multiplayer, framerates are locked to 60FPS to ensure a consistent and competitive online racing experience for all players across a variety of PC hardware configurations and the Xbox Series X|S consoles."

Racing against other players is locked to 60 so you getting 60 anyway is good then.
 
saor said:
Building a pc for rendering & cad. Don't need anything super high end, the specs here are adequate. CPU & GPU - am I missing something that's going to provide a big difference if I spend +- R1000 more? Any issues with that mobo? That PSU ok for the GPU>

View attachment 1601696
mobo: https://www.wootware.co.za/asus-rog...-b550-socket-am4-atx-desktop-motherboard.html
Click to expand...
I wouldn't skimp on the PSU like that. 550W may be a little tight for the specs under a decent load. Remember, 550W is max output, not necessarily output it can comfortably sustain. And that doesn't take into account transient spikes or in-rush.
I have a 7800X3D and 4070Ti on a 650W which is fine, so you should be able to get away with that. But I would definitely upgrade to a better, higher wattage PSU.

Also, not sure how many fans that case has as standard but you may want to grab an additional 1 or 2 for decent airflow (depending how many it has as standard.)
 
saor said:
Building a pc for rendering & cad. Don't need anything super high end, the specs here are adequate. CPU & GPU - am I missing something that's going to provide a big difference if I spend +- R1000 more? Any issues with that mobo? That PSU ok for the GPU>

View attachment 1601696
mobo: https://www.wootware.co.za/asus-rog...-b550-socket-am4-atx-desktop-motherboard.html
Click to expand...

You can get 4 extra cores for the same price with a 5900x...scaling in productivity is generally 1:1 with core count for cpu's in the same gen so that should be 50% faster in tile-based rendering. Rendering doesn't make much use of the cache or clock speed advantage in the 5800x3d.

Edit: I forgot the 3d v-cache models don't clock as high as the non-3d models...not that it matters much in productivity. Actually except for benchmarks, a few hundred mhz here and there doesn't matter generally. Architecture and core count matter a lot more.
 
Last edited:
UrBaN963 said:
I wouldn't skimp on the PSU like that.

Also, not sure how many fans that case has as standard but you may want to grab an additional 1 or 2 for decent airflow (depending how many it has as standard.)
Click to expand...
Updated the PSU. Open to any case recommendations if you have. Don't want LED or anything. Black & minimal.
Neoprod said:
You can get 4 extra cores for the same price with a 5900x...scaling in productivity is generally 1:1 with core count for cpu's in the same gen so that should be 50% faster in tile-based rendering. Rendering doesn't make much use of the cache or clock speed advantage in the 5800x3d.
Click to expand...
Thanks that makes more sense.

Updated:

12.png
 
Last edited:
Status
Not open for further replies.
Back
Top
Sign up to the MyBroadband newsletter
X