Overclock.net banner

[Wccftech] AMD Vega 10 3DMark Fire Strike Benchmark Entry Spotted

18K views 208 replies 67 participants last post by  ZealotKi11er 
#1 ·
Quote:
The Vega 10 687F:C1 Engineering Sample 3DMark Fire Strike Performance Result
The 3DMark Fire Strike entry shows a graphics score of 17801, which is roughly 1400 points more than an R9 Fury X, nearly exactly the same as a Maxwell GTX Titan X and less than a couple hundred points below a GTX 1070.Reference

Putting The Vega 10 687F:C1 Performance Figures Into Perspective
Because this is an engineering sample we have to put its performance into perspective of what AMD expects its final production ready Vega 10 graphics cards to be able to deliver. Thanks to AMD's DOOM demo we can seamlessly connect the dots. We know that the Vega 10 graphics card that AMD has been using all this time in its demos over the past several months was in fact the same 687F:C1 prototype running at 1.2GHz.

1200MHz is a fairly modest clock speed, even when compared to Polaris 10. We know the tweaked version of Polaris 10, Polaris 20, which is now in every RX 580 graphics card out there runs at 1350MHz+. In fact many RX 580 graphics cards widely available today run at 1400MHz+ right out of the box.

As with any semiconductor product, between the very first 687F:C1 Vega engineering graphics board and the final production ready Radeon RX Vega are going to be many iterations of the silicon. Engineering teams work tirelessly during product bring-up to produce one iteration after the other as they optimize the silicon and iron out issues.
During this process and as the product matures overtime, more and more performance is squeezed out of the silicon before it's finally ready to ship. This is why we always see the final products, whether it be AMD, Nvidia or even Intel chips, end up running at significantly higher clock speeds and considerably higher efficiency than their engineering sample counterparts.
http://wccftech.com/amd-vega-10-3dmark-firestrike-benchmark-entry-spotted-687fc1-device-8gb-700mhz-vram-1200mhz-core-clock/

WCCF article. Make of it what you will...
 
#4 ·
Remember the 290x rebrand to the 390x? Well it is happening exactly the same again with FuryX to Vega.
Or maybe trolling.
rolleyes.gif
 
  • Rep+
Reactions: iLeakStuff
#5 ·
Quote:
Originally Posted by rluker5 View Post

Remember the 290x rebrand to the 390x? Well it is happening exactly the same again with FuryX to Vega.
Or maybe trolling.
rolleyes.gif
Fury X scores about 15500 in Firestrike. With a 1050MHz clock.
This Vega chip runs at 1200MHz and scores 17800 so yeah lol it is almost like it was a overclocked Fury X with 4GB more VRAM.

This surely must be the smallest Vega right? No way they release a full Vega barely any faster than Fury X....
tongue.gif
 
#6 ·
Quote:
Originally Posted by iLeakStuff View Post

So that Vega scores 17800 in Graphic score


Comparison

Wonder if C1 is the smallest Vega or the biggest. There is a C3 as well but just because it got a higher number doesnt mean its bigger
thinking.gif


Oh and link to 3DM score
http://www.3dmark.com/fs/12296284
I was just watching LTT's podcast. He said they had just went to an event, and that he thought vega was much better than expected.

I fully expect C1 to be the lowest-clocked of the three Vega 10 SKU's, probably with 20% SP's disabled, slower HBM, meant to be an interim competitor to Nvidia's midrange (1070/1080) products as they clearly can't just pump out Vega 11 with their limited funds.
 
#7 ·
Quote:
Originally Posted by Buris View Post

I was just watching LTT's podcast. He said they had just went to an event, and that he thought vega was much better than expected.

I fully expect C1 to be the lowest-clocked of the three Vega 10 SKU's, probably with 20% SP's disabled, slower HBM, meant to be an interim competitor to Nvidia's midrange (1070/1080) products as they clearly can't just pump out Vega 11 with their limited funds.
Leaked slides from that ROCm conference showed Vega 11 is due this year to replace Polaris. I doubt they'd release an interim product so close (<1yr) to a new product that fulfills the spot perfectly.
 
#8 ·
Quote:
Originally Posted by Buris View Post

I was just watching LTT's podcast. He said they had just went to an event, and that he thought vega was much better than expected.

I fully expect C1 to be the lowest-clocked of the three Vega 10 SKU's, probably with 20% SP's disabled, slower HBM, meant to be an interim competitor to Nvidia's midrange (1070/1080) products as they clearly can't just pump out Vega 11 with their limited funds.
I really hope so
 
#9 ·
Quote:
Originally Posted by geoxile View Post

Leaked slides from that ROCm conference showed Vega 11 is due this year to replace Polaris. I doubt they'd release an interim product so close (<1yr) to a new product that fulfills the spot perfectly.
I disagree. Vega 11 will be nearly half the size of Vega 10, will give around the mid-7Tflop range as far as perf. goes. This is a cut down Vega 10, with 30% slower memory, 20-30% less cores, and will be clocked significantly higher than what's being tested (probably 1400mhz) and will land in between the 1070 and 1080.

Full Vega 10 will land in between the 1080 and 1080 Ti, perhaps beating the Titan Xp in a select few DX12/Vulkan titles.

Vega 11 will more or less land right around the 1070 level (losing in DX11, winning in DX12/vulkan) and will completely replace the existing P10 lineup.

(As far as the top of the line it will probably just be another watercooled-monstrosity, binned, highest available clocks already at stock- probably competing with Titan Xp but losing when the titan is OC'ed)
 
#10 ·
Imagine if this is a cut down Vega 10...

It would be quite interesting if a cut down Vega competes with a GTX 1080.
 
#13 ·
Quote:
Originally Posted by Buris View Post

I disagree. Vega 11 will be nearly half the size of Vega 10, will give around the mid-7Tflop range as far as perf. goes. This is a cut down Vega 10, with 30% slower memory, 20-30% less cores, and will be clocked significantly higher than what's being tested (probably 1400mhz) and will land in between the 1070 and 1080.

Full Vega 10 will land in between the 1080 and 1080 Ti, perhaps beating the Titan Xp in a select few DX12/Vulkan titles.

Vega 11 will more or less land right around the 1070 level (losing in DX11, winning in DX12/vulkan) and will completely replace the existing P10 lineup.

(As far as the top of the line it will probably just be another watercooled-monstrosity, binned, highest available clocks already at stock- probably competing with Titan Xp but losing when the titan is OC'ed)
I see now that this is wishful thinking, because you're basically dreaming of a full Vega that has 43% more cores (if a cutdown chip has 30% less) and is clocked 30% higher in core frequency. Your dream Vega 10 would demolish a 1080TI but that sounds like fantasy to me.

Unfortunately AMD isn't going to cut down a monolithic 500mm chip by 30% for a mainstream product.
 
#15 ·
That's funny they compared it to a Titan X Maxwell, when I saw the earlier leaked Timespy score, I noticed my 980Ti beat it and thought how is Vega gonna compete with Pascal if it's struggling to beat Maxwell. I still think Vega will do better with final drivers.
 
#16 ·
Quote:
Originally Posted by iLeakStuff View Post

Fury X scores about 15500 in Firestrike. With a 1050MHz clock.
This Vega chip runs at 1200MHz and scores 17800 so yeah lol it is almost like it was a overclocked Fury X with 4GB more VRAM.

This surely must be the smallest Vega right? No way they release a full Vega barely any faster than Fury X....
tongue.gif
If it is the true Vega, this is a biggest flop in RTG, after 2 years, they rebrand the same Fury X with 10% higher speed lol.

This cut down vega need to be no more than 3072 shaders with 1.2GHz, in order for the full flat vega of 4096 at 1.5GHz to be on par with 1080 Ti. That is just too much hope to ask for considering scaling of cores + clock is not directly linear.
 
#17 ·
Quote:
Originally Posted by geoxile View Post

I see now that this is wishful thinking, because you're basically dreaming of a full Vega that has 43% more cores (if a cutdown chip has 30% less) and is clocked 30% higher in core frequency. Your dream Vega 10 would demolish a 1080TI but that sounds like fantasy to me.

Unfortunately AMD isn't going to cut down a monolithic 500mm chip by 30% for a mainstream product.
The best wish is a 25% cut down from FULL Vega and is clocked 20% less, which means the actual release of FULL VEGA is 67% faster at launch and that is the only way to be on par with 1080 Ti.
 
#19 ·
If this true AMD is sunk in the high end graphics dept forever IMO. Here is a comparison of my sig rig vs the linked FS score http://www.3dmark.com/compare/fs/12296284/fs/9351182
I'm crushing that system with a highly OCed 980ti & decently OCed 6700k, it has a friggin !800X in it, it should almost match me with a RX 580/1060 because of the CPU. I really hope this is just all BS cause even if this is the lowest, most cut down Vega variant it is still very bad news for those of us who are trying to wait it out to see what Vega brings.
thumbsdownsmileyanim.gif
 
#22 ·
I'm not one for speculation but there is simply no way AMD would release a new flagship card that only equals performance clock for clock and core for core compared to a card they released two years ago.

My Fury X @ 1225 core and stock 500 Mhz HBM scores 18,296 vs this supposed Vega at just 25 Mhz slower on the core but +200 on the HBM is only able to manage 17,801. I really doubt it.

http://www.3dmark.com/compare/fs/12296284/fs/11329395

To think that they added no tweaks from Fiji to Vega makes simply makes no sense. Look at the improvements made to Polaris from Tonga. I think the performance increase was somewhere around 20%, core for core and clock for clock.

At a minimum, one should at least expect that same 20% increase from Fiji to Vega, clock for clock and core for core.
 
#23 ·
Quote:
Originally Posted by Ish416 View Post

I'm not one for speculation but there is simply no way AMD would release a new flagship card that only equals performance clock for clock and core for core compared to a card they released two years ago.

My Fury X @ 1225 core and stock 500 Mhz HBM scores 18,296 vs this supposed Vega at just 25 Mhz slower on the core but +200 on the HBM is only able to manage 17,801. I really doubt it.

http://www.3dmark.com/compare/fs/12296284/fs/11329395

To think that they added no tweaks from Fiji to Vega makes simply makes no sense. Look at the improvements made to Polaris from Tonga. I think the performance increase was somewhere around 20%, core for core and clock for clock.

At a minimum, one should at least expect that same 20% increase from Fiji to Vega, clock for clock and core for core.
Vega's bandwidth is actually less than your Fiji chip because the HBM bus is thinner (2048-bit on Vega vs 4096-bit on Fiji).

On the other hand, Polaris was only about 7% faster with equal shaders, clocks and memory bandwidth against Tonga, but Vega is supposed to be a major arch update, while Polaris merely added a few refinements.

I'm really hoping that AMD is intentionally holding back the performance of Vega's engineering samples. I think there is a good chance of that. At the very least, Vega designs should hit clocks around 1500MHz.
 
#24 ·
Quote:
Originally Posted by Rob27shred View Post

If this true AMD is sunk in the high end graphics dept forever IMO. Here is a comparison of my sig rig vs the linked FS score http://www.3dmark.com/compare/fs/12296284/fs/9351182
I'm crushing that system with a highly OCed 980ti & decently OCed 6700k, it has a friggin !800X in it, it should almost match me with a RX 580/1060 because of the CPU. I really hope this is just all BS cause even if this is the lowest, most cut down Vega variant it is still very bad news for those of us who are trying to wait it out to see what Vega brings.
thumbsdownsmileyanim.gif
If it is the most cut down version it isn't bad news. Your highly overclocked 980ti is like a 1070 (or a little better), one of the cards AMD needs to compete against. They are missing three cards from their lineup, one to compete with the 1070, one for the 1080 and one for the 1080ti. I know they have older cards that compete fine against the 1070 but they need a new one. They could probably get away with ignoring the 1080 and compete with the 1070 and 1080ti, especially if they undercut nvidias pricing a bit.
 
#25 ·
If this is :C1 it's the same one from the demo, it's probably the full version.

But it's running slower than what the production release version will be (~1500Mhz), so you have to keep that in mind.

Also, I'm sure they're working on optimizing the drivers for current games and benchmarks even as we speak.

I just wish we'd get a real leaked benchmark (or review, like the French magazine with Ryzen) to end all this speculation
smile.gif
 
#26 ·
Also remember Pascal isn't that big of a leap past Maxwell, and NV released it. Of course, they were able to massively scale up the clocks to make up for that
wink.gif


I seem to remember reading somewhere that Pascal is actually a tad slower clock for clock.

But, hard to argue with 2100Mhz OC's
wink.gif
 
This is an older thread, you may not receive a response, and could be reviving an old thread. Please consider creating a new thread.
Top