There was a poster somewhere on this forum - possibly in this monstrous thread - that wished ardently for AMD GPUs to gain some special operating capabilities while running in Ryzen systems. Perhaps he has gotten wish . . . sort of.
The prefetch and prediction happens every cycle on the CPU, and there are 4 billion of them a second.as understand zen should perform better if you run application several times.
https://thetechaltar.com/amd-ryzen-1800x-performance/2/
For those who looked for results from an AMD GPU comparing Intel and AMD. At the bottom of the page is an RX 480 tested in 720p (by my request )
It was tested against a 6900K (as that is the performance benchmark in my opinion, not 7700K). Ryzen fairs much better here, actually outperforming the 6900K in most games with one expection being shadow of mordor.
There's also some 4+0 vs 2+2 comparison there, for those interested in more data points.
Was the 6900k at stock? I have to imagine that the 1800x had a clockspeed advantage here. It really shouldn't be beating the 6900k otherwise.https://thetechaltar.com/amd-ryzen-1800x-performance/2/
For those who looked for results from an AMD GPU comparing Intel and AMD. At the bottom of the page is an RX 480 tested in 720p (by my request )
It was tested against a 6900K (as that is the performance benchmark in my opinion, not 7700K). Ryzen fairs much better here, actually outperforming the 6900K in most games with one expection being shadow of mordor.
There's also some 4+0 vs 2+2 comparison there, for those interested in more data points.
I'd need to ask him about TB3, but the AdoredTV thing is not what we set out to test. It was more in general to see if AMD GPU's play nicer with Ryzen. The threaded optimization tests later in the review were done to find the cause of NVIDIA GPU's not liking Ryzen as much, but it didn't answer the question.Was Turbo 3.0 enabled?
Anyway the main idea would be to test DX11 vs DX12 , Radeon vs Nvidia, 4 cores vs 8 cores (preferably 8 cores from both Intel and AMD) since Nvidia doesn't seem to like many cores under DX12.
Everything stock in this test.Was the 6900k at stock? I have to imagine that the 1800x had a clockspeed advantage here. It really shouldn't be beating the 6900k otherwise.
I am sure there are gains from using AMD cards, but it shouldn't be that large.
Curious numbers, though it has to be said that at least AotS and RotR numbers look suspiciously low for CPU limited scenario.https://thetechaltar.com/amd-ryzen-1800x-performance/2/
For those who looked for results from an AMD GPU comparing Intel and AMD. At the bottom of the page is an RX 480 tested in 720p (by my request )
It was tested against a 6900K (as that is the performance benchmark in my opinion, not 7700K). Ryzen fairs much better here, actually outperforming the 6900K in most games with one expection being shadow of mordor.
There's also some 4+0 vs 2+2 comparison there, for those interested in more data points.
For someone who prefers not to listen to someone with worse success rate on speculations than even myself, what theory?Compare the scores with theory that AdoredTV has came up in last two videos
Here's another interesting video related to the performance differences we're seeing: https://www.youtube.com/watch?v=nIoZB-cnjc0
The take away is that it's likely due to NVidia's use of a software scheduler which works great for DX11 and cases where a single thread largely dictates performance scaling. However, it doesn't scale particularly well beyond 2-4 cores and in scenarios where a game can max out most or all cores is a performance hit because running a software scheduler as part of the driver has some added cost that a hardware scheduler avoids.
It's a really interesting video and worth a full watch even though it clocks in at around 20 minutes.
For someone who prefers not to listen to someone with worse success rate on speculations than even myself, what theory?
I helped with testing methodology. GPU utilization in AoTS and RoTR was pegged far below 100%, so it was CPU limited.Curious numbers, though it has to be said that at least AotS and RotR numbers look suspiciously low for CPU limited scenario.
To be honest, I don't see how our tests factor into Adored's theory. Not that I disagree with him or anything, but I don't see how they at all support or contradict his results and theories. Same for NerdTechGasm.If you want to know what theories, watch the videos. They are on his youtube channel, and the numbers from his test are in line with other scores.
Every test so far which I have seen, using Nvidia GPU put Ryzen 7 below 6900K, regardless of details, game, and resolution.To be honest, I don't see how our tests factor into Adored's theory. Not that I disagree with him or anything, but I don't see how they at all support or contradict his results and theories. Same for NerdTechGasm.
That is a good video. The ending kind of pisses you off though. I never realized it was so easy for Nvidia to handicap AMD's performance.Digital foundry tested RoTR with a 295x2 and TitanXP and basically found the same thing Adored did.
Starts at ~8:15 if the time stamp doesn't work.
https://youtu.be/nLRCK7RfbUg?t=493
Huh, i presume testing sequences were custom runs?I helped with testing methodology. GPU utilization in AoTS and RoTR was pegged far below 100%, so it was CPU limited.
Once again, why would i watch videos of someone with proven track record of malicious incompetency, his fanbase can provide him the money he seeks just fine without my assistance.If you want to know what theories, watch the videos. They are on his youtube channel, and the numbers from his test are in line with other scores.
Just to set record straight, it does not lift Ryzen results anywhere, you have entire video to see for that, except BF1 Dx12... And that's hardly an exception worth considering since it's unplayable.Using AMD GPU with Ryzen CPU lifts framerate higher, or close to Intel CPUs.
I never realized people like making up far-reaching conclusions either.I never realized it was so easy for Nvidia to handicap AMD's performance.
Yah, it's just a huge coincidence that nearly every gameworks title ever made initially runs awful on everyone's hardware, but significantly worse on AMDs.Huh, i presume testing sequences were custom runs?
Once again, why would i watch videos of someone with proven track record of malicious incompetency, his fanbase can provide him the money he seeks just fine without my assistance.
Just to set record straight, it does not lift Ryzen results anywhere, you have entire video to see for that, except BF1 Dx12... And that's hardly an exception worth considering since it's unplayable.
I never realized people like making up far-reaching conclusions either.
Don't remember actually. I'll ask him to check his notes.Huh, i presume testing sequences were custom runs?
No, it is a logical consequence of GCN flaws, nothing else. But i digress, what business nV has hurting it's consumer base? It's not exactly in competition with Ryzen, y'know.Yah, it's just a huge coincidence that nearly every gameworks title ever made runs awful on everyone's hardware, but significantly worse on AMDs.
I mean, logic dictates they have to be, because these look worse than 1080p canned benches on both Ryzen and 6900k. But i'll wait.Don't remember actually. I'll ask him to check his notes.
These tests were done over multiple weeks (starting since before launch), so I don't remember exact details in many areas.
No, it is a logical consequence of GCN flaws, nothing else. But i digress, what business nV has hurting it's consumer base? It's not exactly in competition with Ryzen, y'know.
.
Out of curiosity, mind linking me examples?I mean, logic dictates they have to be, because these look worse than 1080p canned benches on both Ryzen and 6900k. But i'll wait.
So, what Dx11 has to do with bad geometry throughput and general front-end performance, that as such limits access to vast overkill of ALU array GCN cards boast? I guess that's the time the old saying about bad dancer and his balls applies.*DX11 flaws
https://www.computerbase.de/2017-03.../4/#diagramm-rise-of-the-tomb-raider-dx12-fpsOut of curiosity, mind linking me examples?