New The PC Build Thread

The new Thermalright air cooler benchmarks appear to be in progress. Tom's Hardware teased it a bit, and it seems that Noctua bros might be a bit upset.

Deepcool which is now outlawed in the US has introduced the FT12/14. Fan. Allegedly it improves their coolers quite a bit, and the latest generation coolers will be bundled with it. I am not sure which coolers will launch with it, but I assume the Assassin and AK series. I doubt we will have reviews on these soon. It is a bit expensive though, so I won't replace the existing F120's (unless you want to move them to the case).

I do wish that local sellers could bring Thermalright in. You can buy them on Amazon, and the top-range cost about the same as an ID-Cooling A720 locally.
 
For new PC builders, PSU makers are abandoning the 80 Plus CLEAResult ratings. It is not a bad thing. Corsair announced it publicly, but others have also adopted Cybenetics ratings and will leave 80 Plus.

Corsair published an explainer where they in some layman language explain why Cybernetics is better to use:


Plus it will save them a buck.
 
Lol,

Screenshot 2024-08-14 175606.png

I don't see why anyone with a 7000 series would NEED to upgrade to a 9000 series. I did a write-up on Reddit, but not going to link to that, lol. In short, I don't expect the new X3D CPUs to be a game changer, unless there is an architectural change to the V-Cache, which in my opinion is unlikely. For new AM5 users, this is a good upgrade, but 7000 series CPUs are being discounted.

AMD has also updated their review notes, a bit late, but something tells me...


AMD set to boost TDP for Ryzen 5 9600X and Ryzen 7 9700X to 105W with upcoming AGESA update

For the compute industries, these CPUs will be welcomed, but in eco-mode.
 
Lol,

View attachment 1751089

I don't see why anyone with a 7000 series would NEED to upgrade to a 9000 series. I did a write-up on Reddit, but not going to link to that, lol. In short, I don't expect the new X3D CPUs to be a game changer, unless there is an architectural change to the V-Cache, which in my opinion is unlikely. For new AM5 users, this is a good upgrade, but 7000 series CPUs are being discounted.

AMD has also updated their review notes, a bit late, but something tells me...




For the compute industries, these CPUs will be welcomed, but in eco-mode.

People still watch LTT stuff? the guy doesnt even know the difference between raid 5 and raid 1 without googling it
 
Mixed results continue to plague the benchmarks.


Not all bench systems are at the same spec, neither are 'optimisations' or otherwise BIOS settings amongst Windows settings. Nobody who benched is attempting to patch scheduling issues with Process Lasso (which is a hit-and-miss). Reviewers collaborate using notes with the review samples. Odd situation. There is no real standard. Then you have things like the silicon lottery, among other things like this:


Just see the Phoronix benchmarks they did with the 9600X/9700X on Linux. It is all unique.

Maybe the secret sauce is in the memory :unsure:
 
Lol,

View attachment 1751089

I don't see why anyone with a 7000 series would NEED to upgrade to a 9000 series. I did a write-up on Reddit, but not going to link to that, lol. In short, I don't expect the new X3D CPUs to be a game changer, unless there is an architectural change to the V-Cache, which in my opinion is unlikely. For new AM5 users, this is a good upgrade, but 7000 series CPUs are being discounted.

AMD has also updated their review notes, a bit late, but something tells me...




For the compute industries, these CPUs will be welcomed, but in eco-mode.

To add to the above:

 
The Phoronix benchmarks are also out,


As can be seen, the 9000 series obliterates everything else. Were the other CPUs retested? I won't know. They say so:

All of these processors were tested on Ubuntu 24.04 LTS with the Linux 6.10 kernel and GCC 13.2 default compiler. All motherboards were with their latest BIOS, 2TB Corsair MP700 PRO NVMe SSD, and Radeon RX 7900 GRE graphics used throughout. All DDR5-capable processors were using 2 x 16GB DDR5-6000 memory while the Ryzen 5000 series was using 2 x 16GB DDR4-3600 memory.

Parking/scheduling is weird.

Here is Wendell's review too:

 
AMD Ryzen 5 5500X3D listing hints at most affordable 3D V-Cache CPU yet — socket AM4 rides again

AMD might be prepping another mainstream processor with 3D V-Cache, originally a premium feature aimed at gamers, if a new listing in the Eurasian Economic Union customs database is accurate. AMD's Ryzen 5 5500X3D discovered by @harukaze5719 promises to be the cheapest X3D processor from the company that will offer great single-thread performance, something that gamers want.

The listing in the EEU customers database does not guarantee that the product listed will ever be launched, so take the information about the Ryzen 5 5500X3D with a grain of salt for now. But truth be told, this is not the first mention of this CPU as blogger chi11eddog predicted the launch of a six-core Ryzen 5 5500X3D in late November 2023.


AMD's six-core Ryzen 5 5500X3D with 96 MB of L3 cache reportedly operates at 3.0 GHz – 4.0 GHz, so its single-thread performance will be similar to that of eight-core Ryzen 7 5700X3D (which has a boost clock of 4.10 GHz) and generally lower than that of other Ryzen 5000X3D processors with 3D V-Cache that feature boost clocks of 4.40 GHz – 4.50 GHz. However, this processor could be a fine choice for an inexpensive AM4 desktop aimed at gaming and applications that take advantage of ultra-large caches.

The CPU itself isn't as interesting as the idea of having launched 2 (in the last month) and possibly now a third cpu even later that's compatible with motherboards from 7 years ago. The 5800XT and 5900XT are pretty silly and this rumoured one might not happen (but could be interesting for gamers on a budget) but it's pretty nuts that someone with a B350 motherboard from 2017 could walk into a store and buy a compatible cpu off the shelf in 2025.

I don't think that situation has existed before.
 
Last edited:
Lol,

View attachment 1751089

I don't see why anyone with a 7000 series would NEED to upgrade to a 9000 series. I did a write-up on Reddit, but not going to link to that, lol. In short, I don't expect the new X3D CPUs to be a game changer, unless there is an architectural change to the V-Cache, which in my opinion is unlikely. For new AM5 users, this is a good upgrade, but 7000 series CPUs are being discounted.

AMD has also updated their review notes, a bit late, but something tells me...




For the compute industries, these CPUs will be welcomed, but in eco-mode.

Side note - if you dislike the stupid things 'tubers do in thumbnails, install the DeArrow extension. It replaces Youtube thumbnails with a random screenshot from the video. It also tries to replace the title but it's not good at that - it removes any ALL CAPS SHOUTING but the actual words are usually the same. Still handy if the dumb faces people make gets on your nerves.

Available for Brave and Firefox...not sure about other browsers.
 
A odd thing about TechPowerUp's benchmark is that the 9000 series was benched on BIOS 2204 and the 7000 series was benched on BIOS 2007. This is not explained as to why???

Does the 7000 series not like BIOS 2124 and 2204 (both Combo AM5 PI 1.2.0.0a)? This is on the ROG Crosshair X670E HERO BIOS.
 
A odd thing about TechPowerUp's benchmark is that the 9000 series was benched on BIOS 2204 and the 7000 series was benched on BIOS 2007. This is not explained as to why???

Does the 7000 series not like BIOS 2124 and 2204 (both Combo AM5 PI 1.2.0.0a)? This is on the ROG Crosshair X670E HERO BIOS.

Maybe he ran benchmarks for Ryzen 7000 etc before the BIOS for Ryzen 9000 came out. Go ask Wizzard.

Edit: 7000
 
Last edited:
Maybe he ran benchmarks for Ryzen 5000 before the BIOS for Ryzen 9000 came out. Go ask Wizzard.

There are some odd margins between the reviewer results when comparing Zen 4 to Zen 5. Some reviewers have touched on it, some experienced anomalies. Just listen to Wendell's own experience, but that is only talking to Zen 5.

It is a good processor, no doubt, yes. It is a gaming king, no. What has my attention is the disparities in the results? There must be a cause.

Ignoring Zen, there are oddities with Intel too, but they have their 'known' issues.
 
There are some odd margins between the reviewer results when comparing Zen 4 to Zen 5. Some reviewers have touched on it, some experienced anomalies. Just listen to Wendell's own experience, but that is only talking to Zen 5.

It is a good processor, no doubt, yes. It is a gaming king, no. What has my attention is the disparities in the results? There must be a cause.

Ignoring Zen, there are oddities with Intel too, but they have their 'known' issues.

Haven't watched any reviews since last week so I can't say. Kind of over this launch tbh...nothing to get me excited.

AMD's pricing strategy should be interesting though - AM4 is still alive and kicking, Zen 4 and Zen 5 are much of a muchness so it looks like people will be spoilt for choice. AMD has to recoup wafer costs for Zen 5 but I imagine the server parts are going to do the heavy lifting there and we might still get the usual price drops a few months after launch in the retail space. Where does that put pricing for Zen 4 which has already had some price cuts?

Might be some bargains out there through Black Friday and the December holiday season.
 
The Phoronix benchmarks are also out,


As can be seen, the 9000 series obliterates everything else. Were the other CPUs retested? I won't know. They say so:



Parking/scheduling is weird.

Here is Wendell's review too:


Damn, 100% performance “boost” from my 5900X to a 7900X in Linux

Maybe in a year I’ll do an “upgrade”, but also things still seem pretty instant to me with my workloads (Rider, IntelliJ, etc, docker, browser).
Honestly memory (32gb) is my only “bottleneck” and this is where Linux kind of fails as a desktop, compared to Windows. Windows’ memory scheduling (if that’s what it is called) just feels far superior. On Linux it is easy to run out of memory - then you either need to use EarlyOOM, and have things automatically killed to save the system, or experience lockups until items are dumped.
This never seems to happen on Windows.
But Linux the performance and productivity boost outweighs everything. I’ve just learnt to not open 8 instances of Rider/Webstorm + 5 brave/chrome windows with countless tabs :laugh:
 
Damn, 100% performance “boost” from my 5900X to a 7900X in Linux

Maybe in a year I’ll do an “upgrade”, but also things still seem pretty instant to me with my workloads (Rider, IntelliJ, etc, docker, browser).
Honestly memory (32gb) is my only “bottleneck” and this is where Linux kind of fails as a desktop, compared to Windows. Windows’ memory scheduling (if that’s what it is called) just feels far superior. On Linux it is easy to run out of memory - then you either need to use EarlyOOM, and have things automatically killed to save the system, or experience lockups until items are dumped.
This never seems to happen on Windows.
But Linux the performance and productivity boost outweighs everything. I’ve just learnt to not open 8 instances of Rider/Webstorm + 5 brave/chrome windows with countless tabs :laugh:

I wish I could buy a 9700X system now to expand my home lab, but alas, no budget to do so.
 
This German reviewer seems to have put the record straight, Google translated:

Old Windows and software problems re-released​

With the two large Ryzen 9000 processors, AMD also has further guidelines on how to achieve the best performance. While the first ones are quite simple and concern the energy saving plan, the activated Windows Game Mode and the activated Xbox Game Bar, things get interesting after that. Because AMD is plagued by problems here that have already been present in the past and which came to light especially at the X3D launch last year.

AMD now explains that regular Ryzen 9000s are also plagued by core provisioning. Switching back from a Ryzen 9 9950X to a Ryzen 7 9700X should therefore be avoided. Reinstalling the chipset driver may work, but it doesn't have to. This normal method is only described as a workaround, but in the end it always comes down to a fresh Windows.

"Unfortunately, sometimes Windows does not apply the correct provisioning after the CPU installed has changed. You can try uninstalling then re-installing the AMD Chipset Driver as a workaround, but a fresh install of Windows is ideal. Please make sure to restart then wait for the system to idle so that the provisioning installation may complete.

For the review purposes we highly recommend performing a clean Windows installation to ensure the provisioning is applied correctly. It may save time to have a separate SSD/image for the 9950X/9900X, and 9700X/9600X, so you don't have to reinstall windows every time you switch between a single CCD and dual CCD configuration while benchmarking. We recommend using Cyberpunk as a sanity check. Using a 4090 you should be getting ~210FPS with the 9950X and ~200FPS with the 9900X.

— AMD"

Testers are now slowly approaching the point where almost every second new AMD CPU will have to have a separate SSD in order to avoid constantly losing time in new installations. This is because the X3D CPUs already wanted their own Windows installations with one or two CCDs and, if possible, should not be mixed with regular Ryzen 7000s. With the 9000 generation, this already affects normal CPUs, and with the following X3D variants, the game will continue under exactly the same conditions.

Effects in the ComputerBase test​

The ComputerBase editorial team therefore had no choice but to reinstall Windows and then reinstall it again between the Ryzen 9000 CPUs. The testers did not think the result of the AMD Ryzen 9 9900X was entirely kosher even without AMD's information three working days before the NDA case. The Cyberpunk mentioned crashed, especially in the percentiles, which looked better again with a new, clean Windows, which replaced the five-day-old Windows. In terms of pure FPS, there was no such drastic gain in Cyberpunk, but Baldur's Gate 3, for example, won clearly here. Smoothed out over many games, in which often nothing at all happens, it still only makes up around 4 percent. The Ryzen 9000 processor will of course not be a high-flyer in gaming.

There were no such significant fluctuations in applications. On average, the CPU ultimately reacts no differently than it would without an updated Windows. All individual results for each game and application can be found on the following pages. The AMD Ryzen 9 9950X was tested directly with the new Windows.

...

That was the Ryzen 9000 launch plus outlook​

So that was it, the launch of the AMD Ryzen 9000 on the desktop. In some places it went as expected, but in many others it didn't. And so the expectations that were met were mixed with some disappointment, garnished with missteps at AMD such as a modest naming scheme and strange, unattainable goals and platform problems. The first tests thus degenerated into question time even among experienced testers about what they had done wrong, only to realize that the manufacturer had slightly overshot the mark.

However, the AMD Ryzen 9000 is not a flop, as YouTubers (have to) shout out. From now on, they deliver the best performance in almost all applications, the single-core IPC is outstanding. If modern features such as AVX-512 are used, the performance goes through the roof. It is also clear that the new cores also need to consume power. Nevertheless, AMD has adjusted the consumption of almost every model downwards compared to its predecessor, three times over the TDP/PPT and even in the flagship model over the PPT. So the bottom line is that the solutions are all more efficient than their predecessors.

I am unsure how AMD got ~210 FPS in Cyberpunk. Did they test at 720p?

According to PC Mag:


We saw some interesting results here, but overall, they were mixed. First, with the testbed PC consuming as much as 599W of power at peak consumption levels in our Adobe Premiere run, the Ryzen 9 9950X showed it can still be quite the power hog. It actually pulled more power than the Ryzen 9 7950X in every test we checked, which was unexpected. At the same time, however, it performed pretty well under Cinebench R23 relative to the competition.

Weirdly, Tom's Hardware reviewer, Thomas, made this quip:


I am beginning to question the competence of motherboard vendors and wonder if AMD and Intel need to sell reference boards to the public again.

Anyhow. Weird.
 
Top
Sign up to the MyBroadband newsletter
X