BTW.
Navi 2560 Shaders @ 1.8Ghz vs 2560 shaders @ 1.9Ghz GTX 1080 vs 4096 Shaders @ 1.5Ghz Vega.
And AMD are just getting started with Navi.
Last edited:
-
Competitor rules
Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.
Intel about to expand into Graphics, AMD on the rise, are the tables turning on nVidia?
- Thread starter humbug
- Start date
BTW.
Navi 2560 Shaders @ 1.8Ghz vs 2560 shaders @ 1.9Ghz GTX 1080 vs 4096 Shaders @ 1.5Ghz Vega.
And AMD are just getting started with Navi.
That he is.
Based on the current Navi's & how they look compared to Vega things look promising.
The gaming IPC is about 25% higher than Pascal, equal to Turing.
Its a great start considering AMD like with CPU's came from nowhere, just goes to show what they are capable of when under the right management, i'll credit Dr Lisa Su with that.
Last edited:
Laura Smith, a name i need to remember, looks like she is Dr Lisa Su's Protégé.
I'm at the point where i'm thinking about AMD GPU's again, i am torn between the RTX 2070 and 5700XT and given that they are roughly the same performance (I don't consider 3% faster, at all) but they are both still a bit too expensive, i'm in the market around April / May when i hope they will come down to about ~£300.
Which ever one reaches that ~£300 mark then is the one i'll buy.
Which ever one reaches that ~£300 mark then is the one i'll buy.
Ouch! that's really bad.... Designing good GPU's isn't so easy after all hey Intel?
I wondered what had happened to intel. Weren't they supposed to be starting to release discrete GPUs to the consumer market by now, and heating up some real competition to AMD/Nvidia?
Guess not
Intel is a nepotistic mess.
As soon a Raja Koduri left AMD Dr Lisa Sue set about his Vega Architecture, she increased the IPC by 53% and reduced the power consumption, they are now in Ryzen 4000 Mobile, in 35 Watt packages those iGPU's are quicker than Nvidia's MX250 discrete GPU's.
I said this while Raja Koduri was still at AMD, Raja Koduri is useless.
60 CU Vega with HBM2 vs RDNA1 40 CU with GDDR6 both 7nm ~200 vs ~300 Watts.
And yes Intel Xe GPU's have the same problem, large, hot, slow.....
I said this while Raja Koduri was still at AMD, Raja Koduri is useless.
60 CU Vega with HBM2 vs RDNA1 40 CU with GDDR6 both 7nm ~200 vs ~300 Watts.
And yes Intel Xe GPU's have the same problem, large, hot, slow.....
https://www.hardwaretimes.com/intel-teases-xe-hp-gpu-die-shot-with-tens-of-billions-of-transistors/
What did i say about Raja's GPU's being large?????
The 2080TI is massive, no getting away from that at 775mm^2 (31mm x 24mm) its mahusive....
That's on 12nm.
For perspective:
An R9 290X is 431mm^2 on 28nm, that is quite a large GPU
A 1080TI is 470mm^2 on 16nm that is a large GPU
Vega 64 is 495mm^2 on 14nm, also a large GPU.
A 5700XT is 251mm^2 on 7nm, that is a mid size GPU, its the same size as an RX 590 on 12nm., similar number of shaders on a smaller node than the 5700 but more stuff in the RDNA architecture.
So... this colossus thing, taking the battery as a size guide is about 65mm x 35mm, (2275mm^2) three times the size of an already huge RTX 2080TI
Intel first Xe demo unit was a 50 Watt discrete GPU taken from Tiger Lake bumped up to that power envelope, it had similar performance to a GTX 1050.
Tiger Lake. https://www.guru3d.com/news-story/i...lake-with-xe-based-gpu-and-thunderbolt-4.html the CPU on that thing we know from Wikipedia is 136mm x 107mm (146mm^2) the Xe GPU on that photo looks about 2.5X the size, about 350mm^2.
Intel have stopped talking about Discrete gaming GPU's, they used to talk about it quite a lot, with great passion. All that noise is gone, now they just talk about specialized GPU workloads, the sort of thing i suspect Nvidia and AMD can do with much smaller GPU's.
We ain't getting no gaming graphics competitors to Nvidia and AMD, they are not our knights in Blue armour, designing good GPU's is not easy, its extremely difficult and the bar of quality mostly driven by Nvidia keeps moving, and its way out of reach of what Intel are capable of, Frankly they can't even keep up with AMD on the CPU side, not when you haven't got infinite cooling and power delivery to brute force your way up, not by a long way....
I do find this extremely frustrating.
What did i say about Raja's GPU's being large?????
The 2080TI is massive, no getting away from that at 775mm^2 (31mm x 24mm) its mahusive....
That's on 12nm.
For perspective:
An R9 290X is 431mm^2 on 28nm, that is quite a large GPU
A 1080TI is 470mm^2 on 16nm that is a large GPU
Vega 64 is 495mm^2 on 14nm, also a large GPU.
A 5700XT is 251mm^2 on 7nm, that is a mid size GPU, its the same size as an RX 590 on 12nm., similar number of shaders on a smaller node than the 5700 but more stuff in the RDNA architecture.
So... this colossus thing, taking the battery as a size guide is about 65mm x 35mm, (2275mm^2) three times the size of an already huge RTX 2080TI
Intel first Xe demo unit was a 50 Watt discrete GPU taken from Tiger Lake bumped up to that power envelope, it had similar performance to a GTX 1050.
Tiger Lake. https://www.guru3d.com/news-story/i...lake-with-xe-based-gpu-and-thunderbolt-4.html the CPU on that thing we know from Wikipedia is 136mm x 107mm (146mm^2) the Xe GPU on that photo looks about 2.5X the size, about 350mm^2.
Intel have stopped talking about Discrete gaming GPU's, they used to talk about it quite a lot, with great passion. All that noise is gone, now they just talk about specialized GPU workloads, the sort of thing i suspect Nvidia and AMD can do with much smaller GPU's.
We ain't getting no gaming graphics competitors to Nvidia and AMD, they are not our knights in Blue armour, designing good GPU's is not easy, its extremely difficult and the bar of quality mostly driven by Nvidia keeps moving, and its way out of reach of what Intel are capable of, Frankly they can't even keep up with AMD on the CPU side, not when you haven't got infinite cooling and power delivery to brute force your way up, not by a long way....
I do find this extremely frustrating.
So you could eat your dinner off this thing....
Thats what he did when he was at AMD too, he done those Vega cards for compute, that he also knocked out as gaming cards.
They were doing these.
lol.....
The 5700XT while not perfect is a good start to AMD's new found purpose, in performance terms its a little beast and there are no caveats to that performance, it deals with high tessellation as well if not better than Nvidia.
The only two thing's that lets it down is AMD are clearly not binning them, if they kept the best silicon for the 5700XT it could be better tuned and draw as little as 150 Watts, which for 2070 <> 2070 Super performance is very good. And some AIB's build them like it was a Friday afternoon and couldn't be bothered. Its only AMD's own exclusive partners that seem to do a good job of it, well... that is except XFX but they have been the exception for some years now. The Powercolor and Sapphire ones are very good, Gigbabite seem to be the exception out of main vendors in that it is also good. The Asus TUF is shockingly bad and MSI's two lower end ones are not much better.
I was going to say there are a couple of features in the Radeon Software that do this thing, not sure what people smoking, maybe its because they dont use an AMD card so assume it doesnt exist as it hasnt got the same name?
Yeah i was going to say the same thing, I have it turned on all the time, i run games at my native res and i have to say for all the complaints i have about AMD drivers, see the AMD Drivers thread.... that works really well, in every game and looks great.