![]-[itman](/forums/data/avatars/m/9/9309.jpg?1650325391){kind=avatar}
Pretty much.
-
Welcome to Overclockers Forums! Join us to reply in threads, receive reduced ads, and to customize your site experience!
It looks like AMD found the missing gaming performance.
- Thread starter wade7575
- Start date
Pretty much.
- Joined
- Mar 7, 2008
Looking at the Hardwareluxx writeup now (link from EarthDog's earlier post). They say (translated):
I also looked at the changes reported for the 7800X3D since I'm debating if I want to install it yet. First number I assume is average fps, 2nd is probably "lows" but I don't see exactly what they measure. I expressed it as percentage change from applying the patch.
Starfield: +3.0 +13.2
CP2077: +10.7 +61.5
F1 2024: +6.2 +9.9
Spider-Man: Miles Morales: +4.9 +3.9
Ratchet & Clank: +3.1 -3.7
BG3: +3.1 -2.5
Control: +0.8 -9.1
Anno 1800: +8.2 +6.2
Starfield was at 1080p, the rest were at 720p and still using DLSS so the render resolution would be miniscule. Gains of up to 10%, while nice if it is free, isn't going to radically change anything. And this is at an absurdly low test resolution. I play at 4k. That'll squash the differences back to zero.
Some titles did see bigger gains in the lows which could be more interesting than the average fps uplift. There were some regressions here too, but again I don't know their test methodology.
Last edited:
Testing like this bothers me. 720p testing is only relevant at 720p (which 0.23% use). It's results are supposed to isolate the cpu in this case, and it does, but it doesn't scale with higher resolutions. So.... what good is that value for everyone else if that's all they report? For example, it's reported there is a 10% increase at 720p, but at 1080p it's 3% (and potentially even less higher). I just don't see the value of the isolation of that number without sharing the rest of the resolutions.
It varies wildly.
Last edited:
- Joined
- Mar 7, 2008
From https://www.mersenneforum.org/node/22559#post1052498
Above post from writer of Y-cruncher, who has done the best technical look at Zen 5 that I'm aware of so far.
I do wonder what behind the scenes stuff is going on between AMD and MS. Was MS expected to have put the update into general release before Zen 5's original launch date?
- Joined
- Oct 16, 2014
- Location
- Allagash, Maine
When can we expect x870e available on Amazon n Newegg?
Some really in-depth technical write-ups on Zen 5 were done by ChipsandCheese and David Huang. I'm linking to C&C's desktop article but they have additional ones covering other Zen 5 variants as well.
AMD’s Ryzen 9950X: Zen 5 on Desktop
AMD’s desktop Zen 5 products, codenamed Granite Ridge, are the latest in the company’s line of high performance consumer offerings. Here, we’ll be looking at AMD’s Ryzen 9 9…
chipsandcheese.com
So there's a practical viewpoint (nobody games at that low of resolution) and the more academic viewpoint (want to entirely sure the GPU is out of the equation to compare CPUs, even if it's unrealistic). I see value in both approaches, except when reviewers do low res + low settings. In almost all games, at least a few of the settings depend on the CPU and turning those off/down makes the comparison nearly pointless. This basically just turns it into a draw call test, at which point you might as well just run a synthetic draw call test.
- Thread Starter
- #28
The ASUS ROG SRIX is supposed to be out Sept 30th and there's no word on when the rest come out which is really stupid if you ask me,they should released 1/3 to 1/2 on the same day and some gaming and some Workload mobo's.
https://www.guru3d.com/story/asus-r...oxing-x870e-to-be-released-september-30-2024/
Most models are expected no earlier than mid-September. I assume the distribution already has many of them, but because of internal agreements, they can't sell them yet. Some stores may sell them earlier.
I don't expect anything exceptional from these motherboards. The main difference is USB4, which higher 600 series motherboards already have.
I guess we will see more new things from the new Intel series. I was counting on various changes in the new AMD, but we have to wait some more for something really new.
@wade7575
Why do you want the ProArt series? It looks pretty good, but I assume it's more like TUF with different covers, so it's a nice looking mid shelf that won't have as good support (BIOS etc.) as Strix or higher series. I can be wrong, but this is how ASUS has worked for years. On the other hand, there are so few changes that I doubt these motherboards will need many BIOS updates, and I don't think we will see any higher RAM too as all mobos are specified to be about the same as X/B 600 series. If I'm right, AMD also won't support CUDIMM (at least not at the start).
I don't expect anything exceptional from these motherboards. The main difference is USB4, which higher 600 series motherboards already have.
I guess we will see more new things from the new Intel series. I was counting on various changes in the new AMD, but we have to wait some more for something really new.
@wade7575
Why do you want the ProArt series? It looks pretty good, but I assume it's more like TUF with different covers, so it's a nice looking mid shelf that won't have as good support (BIOS etc.) as Strix or higher series. I can be wrong, but this is how ASUS has worked for years. On the other hand, there are so few changes that I doubt these motherboards will need many BIOS updates, and I don't think we will see any higher RAM too as all mobos are specified to be about the same as X/B 600 series. If I'm right, AMD also won't support CUDIMM (at least not at the start).
You'll see most x870 boards on 9/30... no earlier... at least reviews that show performance of these boards aren't allowed until 9/30. I'd expect most to release that day.
In this case, however, it takes an unrealistic situation in order to show/magnify a difference. So while it's neat to see academically, it doesn't do much in the end considering it doesn't scale and 'nobody' plays there.
Again, cool to see that data point, but, that's as far as I would take it. It's downright misleading offering that value only.
I understand the purpose.
In this case, however, it takes an unrealistic situation in order to show/magnify a difference. So while it's neat to see academically, it doesn't do much in the end considering it doesn't scale and 'nobody' plays there.
Again, cool to see that data point, but, that's as far as I would take it. It's downright misleading offering that value only.
Last edited:
- Joined
- May 29, 2005
Yep agreed with the 720p piece. It's like measuring a Ferrari, McClaren, Bugatti, and Lamborghini and how quickly they can get to 100MPH on a neighborhood street. It tells you how fast they can get there in a never used use -case.
At least HWU uses 1080p, which is still a dominant resolution. But as I game at 4K I imagine the difference is minimal at best, but would be nice to see those numbers anyway in case it does make a difference. Although I think we generally agree that having a faster CPU just feels like opening more lanes on the racetrack for your graphics card to be able to stretch its legs
At least HWU uses 1080p, which is still a dominant resolution. But as I game at 4K I imagine the difference is minimal at best, but would be nice to see those numbers anyway in case it does make a difference. Although I think we generally agree that having a faster CPU just feels like opening more lanes on the racetrack for your graphics card to be able to stretch its legs
The free performance gain is always good. Free like no one expected that, even though if someone made it right, we could see it a long time ago. It feels like all were happy with the performance before this update, so it's just a cherry on top for all users.
Numbers seem to be mainly for marketing. 10 or 20% counts almost only when we miss some FPS. I doubt that most people with 100FPS+ care if they have ~15% more or less FPS. Most don't even know how many FPS they have and as long as it's smooth then it's fine.
It feels weird to see the market pushing 4K and at the same time trying to prove that something is so much better showing 720p/1080p results.
Numbers seem to be mainly for marketing. 10 or 20% counts almost only when we miss some FPS. I doubt that most people with 100FPS+ care if they have ~15% more or less FPS. Most don't even know how many FPS they have and as long as it's smooth then it's fine.
It feels weird to see the market pushing 4K and at the same time trying to prove that something is so much better showing 720p/1080p results.
It's marketing/for clicks... flavor of the week. The week before....psssssssssh!
- Joined
- Oct 16, 2014
- Location
- Allagash, Maine
I can not over state how much work went into tweaking around apps n performance to get native 4k/24hz video playback on a Windows OS .
Hindsight makes it seem easy , tho a decade ago...
720p is not a valuable statistic from this perspective.
Retro Gaming?
- Thread Starter
- #37
The Strix has les PCIe slots then the ProArt and I need at least 2 for sure to run the stuff I want to run.
- Thread Starter
- #38
If you guys look at this video at the 2:22 point you can see where he mention's about VBS being turned off when it should be on for best performance.
He mentions VBS be turned by default in previous builds of Windows 11 and I'm not sure if it is turned off by default now when just the patch is used or if it is also turned off when you do a fresh install with an updated ISO.
He's say' you need to check and make sure VBS turned on to get the best performance.
In this video he talks about another update for AMD regarding the latency problems that affect the CCX and CCD's,from what I gathered he say's that with Zen 4 the latency was around 18 to 20ns( nano seconds ) and that in Zen 5 for the very same thing it is 180ns
He mentions VBS be turned by default in previous builds of Windows 11 and I'm not sure if it is turned off by default now when just the patch is used or if it is also turned off when you do a fresh install with an updated ISO.
He's say' you need to check and make sure VBS turned on to get the best performance.
In this video he talks about another update for AMD regarding the latency problems that affect the CCX and CCD's,from what I gathered he say's that with Zen 4 the latency was around 18 to 20ns( nano seconds ) and that in Zen 5 for the very same thing it is 180ns
7950x was ~76ns.... 9950x is ~180+...
- Joined
- Oct 16, 2014
- Location
- Allagash, Maine
Does a bandwidth comparison accompany these core to core latency stats?
Similar threads
