Rumors are that Vega matches Pascal on a perf/TFLOP basis so I would be disappointed if Vega 10 does not at least trade blows with an 1080 Ti. At best I am expecting Vega 10 to match 1080 Ti aftermarket cards and at worst I am expecting it to lost to a 1080 Ti FE by 10%.AMD explicitly stated that Vega is designed for both increased clockspeed and increased IPC in the Vega architecture preview that they gave to journalists at the start of the year.
Way I see it, AMD will have failed miserably if this thing doesn't at least match the 1080Ti. Since we already know how many TFLOPs it puts out, that means they need to pull off a large increase in actual performance per FLOP. Their great execution with Ryzen leaves me hopeful that they can pull it off, but the leaks and presentations so far seem underwhelming. Hopefully it's just driver issues or slower engineering sample silicon. Fingers crossed...
Actually with Volta near... AMD Vega must defeat by 15% to 1080 Ti...Rumors are that Vega matches Pascal on a perf/TFLOP basis so I would be disappointed if Vega 10 does not at least trade blows with an 1080 Ti. At best I am expecting Vega 10 to match 1080 Ti aftermarket cards and at worst I am expecting it to lost to a 1080 Ti FE by 10%.
It is not different. Its the same API that both are based on: Mantle. Metal from the start integrated OpenCL for compute, and OpenGL for Graphics in one command, based on low-level API. Vulkan will start soon doing the same thing. There will not be OpenCL, and Vulkan. There will just be Vulkan for Graphics and Compute.I for one have to say I'm very happy AMD got the design win for the iMac Pro. While it might be obvious (Apple have been using AMD dGPUs for quite a while), it'll still be a massive boost in adoption for Radeon Pro cards and software. Now, it's really too bad that Apple won't allow Vulkan on MacOS, but I suppose Metal2 might not be too different in terms of compute, and I can definitely se devs bootcamping Windows on those iMacs. Of course, this will help AMD's bottom line too. Let's just hope it doesn't mean no Vega availability for us (although the miners might see to that already ...).
One thought that I haven't seen brought up is whether or not AMD sacrificed IPC for speed.
So far everyone has assumed at least some minor IPC gains, but what if they traded performance per clock for more speed?
If you take that slide literally, then you have to expect a 2x clock increase as well as 2x IPC. 2x IPC is true for packed math, but obviously not other workloads. 2x clocks (i.e. ~2.7GHz) is not happening. In other words, that slide is, at best, veguely illustrative of "more IPC" and "higher clocks" with no indication whatsoever what either of those mean. It may very well be referring to packed math - or not. That it was mentioned earlier might just mean that that slide is summing things up.You must have missed this slide.
LE: At first I thought this was about increase in IPC because packed math, but then I realized they already talked about it in a previous slide.
Is Vulkan code compatible with Metal? Otherwise, they're not the same, just similar. Of course, the more similar they are the easier it will be to port apps and games, but they'll still need work.It is not different. Its the same API that both are based on: Mantle. Metal from the start integrated OpenCL for compute, and OpenGL for Graphics in one command, based on low-level API. Vulkan will start soon doing the same thing. There will not be OpenCL, and Vulkan. There will just be Vulkan for Graphics and Compute.
Khronos have just did what Nvidia did years ago, but did it in much better fashion. OpenSource.
This post answers your questions.If you take that slide literally, then you have to expect a 2x clock increase as well as 2x IPC. 2x IPC is true for packed math, but obviously not other workloads. 2x clocks (i.e. ~2.7GHz) is not happening. In other words, that slide is, at best, veguely illustrative of "more IPC" and "higher clocks" with no indication whatsoever what either of those mean. It may very well be referring to packed math - or not. That it was mentioned earlier might just mean that that slide is summing things up.
In RX Vega GPU there are 4 Shader Engines that are partitioned into two pieces, with 8 CU's each. It helps with load balancing, efficiency because you can power down unused CU's. Each Compute Cluster of 8 CU's has the same amount of resources available as 16 CUs in Fiji, hence over two times higher Geometry Throughput compared to Fiji.
I'm not asking any questions, I'm simply stating that reading too much into an unlabeled and very vaguely illustrated slide is a poor basis for speculation. If other sources give us other information, let's base our ideas on those sources, and stop going back to this useless slide. Essentially, all it says is "Vega will be better."This post answers your questions.
30-70% are not really the best case. According to PCGH, the 1600X is 45% faster than the similar priced 7600K in terms of applications.He's talking about this slide that's using Cinebench - Ryzen's best case scenario (30-70% performance advantage). The average workload performance comparison in the same price bracket it's much lower, around -3 to 11%.
AMD's slide shows 75% when comparing 1600X to 7600K, much more that 45% that you said.30-70% are not really the best case
According to PCGH, the 1600X is 45% faster than the similar priced 7600K in terms of applications.
He didn't say that 30-70% was the best case for 1600X, he said that Cinebench was the best case for 1600X (and the lineup as a whole). The lineup as a whole has an advantage of 30-70% in Cinebench, but 1600X specifically has an advantage of 70% (of 84% if you go by PCGH's numbers.30-70% are not really the best case. According to PCGH, the 1600X is 45% faster than the similar priced 7600K in terms of applications.
http://www.pcgameshardware.de/CPU-Hardware-154106/Tests/Rangliste-Bestenliste-1143392/
yup. We have a whole forum chock full of ryzen threads just waiting for you guys to have at it (please don't have at it, I'm being cheeky).The CPU debate is kinda derailing guys..
"In the in the Ryzen, Ryzen processor is just as an example, we've launched seven versions of the Ryzen processor to-date and if you look at those seven versions relative to their Intel competitor that's most nearly priced competitor you'll see we're bringing anywhere from 30%, sometimes over 70% performance advantage. And when we launch our Vega graphics processor that's coming very shortly, it will come later this month, you'll see similar level of competitive performance in high-end part of the desktop or high-end part of the graphics market."
https://seekingalpha.com/article/40...-stifel-nicolaus-2017-technology-internet-and
Bold statement by mr. Anderson. Im pretty sure he refers to performance vs price bracket. So lets say that AMD offers 399$ GPU based on RX Vega that is much faster than their competitor. Lets say that 699$ RX Vega GPU is 30% faster than GTX 1080 Ti. This makes sense. SemiAccurate members already jumped to conclusion that it will be 70% faster than GTX 1080 Ti...
This cant be possible. The CEO of nvidia said not to expect any changes in the competitve landscape from Vega.