Well, what if it is not 300W like I have pointed out in this very post?Only 10% in DOOM and 4K as absolute best case. So as expected Vega ends up between 1070 and 1080 in average but using 250-300W.
It might be little bit different, since production process was also new (not to say not ready). So they (GF and AMD) will probably need less time to start full production of Vega chips, due to lessons learned from Polaris projectAnother important part from computerbase:https://www.computerbase.de/2016-12/amd-vega-details/
They say Raja mentioned the Vega samples are just a few weeks old. So anybody who still dreamed of Vega in early Q1 should see, that Vega is really a end of Q2 product. Polaris samples which were shown in december last year were also pretty fresh and then needed more than 6 months to launch. Won't be different with Vega.
Few things are not exactly straight here. In my opinion the MI25 is dual GPU card.
If it's 10% faster than a 10% overclocked 1080 (1900mhz which is a 10% OC), isn't it ~20% faster than a stock 1080?Only 10% in DOOM and 4K as absolute best case. So as expected Vega ends up between 1070 and 1080 in average but using 250-300W.
http://videocardz.com/63700/exclusive-first-details-about-amd-vega10-and-vega20It's FP16 performance, not FP32.
With Vega having the possibility to pack FP16 this explains the 25 TFLOPs.
Only 10% in DOOM and 4K as absolute best case. So as expected Vega ends up between 1070 and 1080 in average but using 250-300W.
Note, however, that first, no Vega-optimized driver was used, but simply a Fiji driver with a little additional debugging work. Second, any view on the graphics card should be denied, so that all the fan slots were taped and there was no airflow in the housing. The graphics card was noticeably noisy and dusted neat hot exhaust air backwards.
About one thing and the Vega MI25 GPU. MI stands for Machine Intelligence. 6 and 8 stands for FP32 performance. So why the hell the GPU - MI25 has 25 there? There are two possibilities. Either it is 300W 12.5 TFLOPs FP32 GPU, or it is dual GPU, with 12.5 TFLOPs of FP32 compute performance, each. There is also no need for that huge package of the single GPU package if it uses HBM memory.
Few things are not exactly straight here. In my opinion the MI25 is dual GPU card.
Well, what if it is not 300W like I have pointed out in this very post?
Also 8 GB of HBM suggest its smaller one of the Vega GPUs.
Yep, with throttling card in completely closed case running Doom in debug mode with Fiji drivers.Only 10% in DOOM and 4K as absolute best case. So as expected Vega ends up between 1070 and 1080 in average but using 250-300W.
If it's 10% faster than a 10% overclocked 1080 (1900mhz which is a 10% OC), isn't it ~20% faster than a stock 1080?
Also, where are the power consumption numbers?
Best case? What? Did you bother to read the rest of the post, the article, or logic at all? What the journalist saw is the worst case performance.
From the article
Vega 10 x2 - 300W TDP. Small Nano Radeon Pro - looks like Videocardz was more correct than they thought initially they will be.
It might be little bit different, since production process was also new (not to say not ready). So they (GF and AMD) will probably need less time to start full production of Vega chips, due to lessons learned from Polaris project
Best case? What? Did you bother to read the rest of the post, the article, or logic at all? What the journalist saw is the worst case performance.
From the article
Single GPU - yes, that is correct. But naming scheme, and technicalities are not in line. MI25 should stand for Machine Learning 25 TFLOPs of FP32.According to the leak, it's a 12.5 Tflop fp32 card that can do 25 Tflop of fp16 calculations.
So how you will the explain potential of 32 GB of HBM?Please read what you post.
They talk about 24 TFLOPs for a single V10 @ 225W ... by this logic MI25 can not be a dual gpu card.
For memory the MI6 has 16GB, MI8 has 4GB, and the MI25 is again unspecified but likely to be 32GB
Yes, I can see the lineup of excuses is long. Not the right drivers, its really a dual card for 300W and so on and on
Vega performs as expected when you look at Polaris.
http://www.anandtech.com/show/10222/nvidia-announces-tesla-p100-accelerator-pascal-power-for-hpc
Note that P100 is also rated @ 300W TDP, as is the MI25. I'm not sure how much the power consumption of a consumer Vega... unless there were DOOM power consumption figures that I've missed ?