Vega hits Compubench

[jasonc_01]

Vega just popped up on Compubench, and it looks to be good but is still Reckt by the new 1080ti.

 

So are we looking at another Radeon line thats competing with Nvidia's former king, the gtx1080 a year to late? 

 

Looks that way, as it seems to fall in to gtx1080 performance territory. 

 

http://www.digitaltrends.com/computing/amd-vega-compute-units-revealed-compubench/

Quote

AMD’s upcoming Radeon RX Vega graphics cards for enthusiasts appear to be nearing their retail release, as a Vega-based GPU appeared on CompuBench. It is listed using the AMD 687F:C1 device ID, the same number that was seen during AMD’s Vega-based Doom Ultra 4K demonstration in January. Thus, there is speculation that the Radeon RX Vega family may arrive in weeks given that near-completed graphics chips typically show up on CompuBench just before their official release.

 

Quote

The benchmark also lists two maximum speeds of 1,000MHz and 1,200MHz, indicating a possible raw computing performance of 8.2 TFLOPS and 9.8 TFLOPS, respectively. By comparison, Nvidia’s GeForce GTX 1070 has a raw computing performance of 6.5 TFLOPS and the GTX 1080 has 8.9 TFLOPS of performance.
 

[MorganO]

rip night theme users kinda

[Bouzoo]

it's not really that hard to remove the formatting. Just check everything and press I_x .

[Orangeator]

4 minutes ago, KOMTechAndGaming said:

rip night theme users kinda

As a multiple year long night theme user, I still haven't got used to these kind of posts. It's the cancer of this forum. Lol.

[DocSwag]

I hate it when they compare TFLOPs... I'm just like... THAT'S NOT HOW IT WORKS!!!

[Droidbot]

12 minutes ago, DocSwag said:

I hate it when they compare TFLOPs... I'm just like... THAT'S NOT HOW IT WORKS!!!

THAT'S NOT A POINT OF COMPARISON, DUMMY

 

AMD FineWineTM Technology will come in clutch and push V E G A over the line

[bakidota]

vega was never designed to beat 1080ti, it was designed to go head to head vs 1080gtx

[vanished]

That's not surprising, at least to me.  I predicted long ago it would be around a 1080, varying (of course) if you're talking about Vulkan or non-Vulkan tests.

[LeStringMan]

3 minutes ago, bakidota said:

vega was never designed to beat 1080ti, it was designed to go head to head vs 1080gtx

Nobody knows where Vega will land, what we do know is that it has 64 compute units, (4096 Stream processors), its the same gpu used in the MI25 which AMD says has 12.5 teraflops, for it to hit 12.5 teraflops it would need a core clock of about 1525 megahertz, we also know that Vega has a lot of new software tricks such as something similar to tile-based rasterization like Nvidia cards now, so the teraflop value is useless compared to previous gen AMD cards, in theory it should be much closer to a 12.5 tflop Nvidia card now, but nobody can say for sure because nobody except AMD knows how efficient their new software tricks are.

[vanished]

Hold on, if the Fury X (same number of shaders) could do 8.6 TFOPS at 1000 MHz, and this new card is gonna do 8.2 TFLOPS at that speed... the IPC has gone down? 

nvm I guess they've changed enough you can't compare them like that

[Mr.Meerkat]

2 minutes ago, Ryan_Vickers said:

Hold on, if the Fury X (same number of shaders) could do 8.6 TFOPS at 1000 MHz, and this new card is gonna do 8.2 TFLOPS at that speed... the IPC has gone down?

Why not eh? Not like IPC matter anyway  

/s 

[LeStringMan]

1 minute ago, Ryan_Vickers said:

Hold on, if the Fury X (same number of shaders) could do 8.6 TFOPS at 1000 MHz, and this new card is gonna do 8.2 TFLOPS at that speed... the IPC has gone down?

Fury X is 1050 mhz which is 8.6 teraflops this card is 8 teraflops at 1000mhz and 9.8 at 1200 mhz. The card in this compubench test is surely downclocked, the RX 480 also appeared on compubench before it launched at a blistering 800mhz iirc.

[vanished]

2 minutes ago, LeStringMan said:

Fury X is 1050 mhz which is 8.6 teraflops this card is 8 teraflops at 1000mhz and 9.8 at 1200 mhz. The card in this compubench test is surely downclocked, the RX 480 also appeared on compubench before it launched at a blistering 800mhz iirc.

I had heard it was 1000 MHz despite what the website said ... interesting; that 50 MHz is enough to make the difference (enough that it means IPC is the same as before)

[DarkBlade2117]

So.. AMD said to share holders they are releasing a "580". If they release a 580 to reach 1070+ levels of performance at the MSRP of a 480 then ok... This is one Vega card of what.. the 2 or 3 different dies from Vega? Doesn't matter how it performs... price is the determining factor.

[Misanthrope]

Oh but wait for:

 

1) DX12

2) drivers 

3) No not THAT DX12 game others like AOTS and....wait for the API to mature and devs to git gud! 

4) No not those drivers. Not those either...it takes like a year for them to mature! 

 

Ah the joys of AMD future proofing. 

[LeStringMan]

Just now, Ryan_Vickers said:

I had heard it was 1000 MHz despite what the website said ... interesting; that 50 MHz is enough to make the difference (enough that it means IPC is the same as before)

Im not sure about ipc in gpus, i do know that Polaris for instance is about 20% faster on average than previous generations at the same clockspeeds (somebody did a test between 380x and 470 clocked identically) There are far more important things like keeping the rasterizers and rops fed (this is where Vega gets most of its changes), as for teraflops that value is pretty much just stream processor count times two times clockspeed, so the Fury X is (4096)(2)(1050)=8,601,600 flops or 8.6 teraflops.

[TechGod]

This could be small Vega. 

 

#JustSayin

[vanished]

Just now, LeStringMan said:

Im not sure about ipc in gpus, i do know that Polaris for instance is about 20% faster on average than previous generations at the same clockspeeds (somebody did a test between 380x and 470 clocked identically) There are far more important things like keeping the rasterizers and rops fed (this is where Vega gets most of its changes), as for teraflops that value is pretty much just stream processor count times two times clockspeed, so the Fury X is (4096)(2)(1050)=8,601,600 flops or 8.6 teraflops.

Oh, totally.  When it comes to determining the actual gaming ability of the card there's a ton of other stuff that we need to factor in.  I was just surprised to see what seemed like a regression in one of the many factors, but it turns out it's equal ¯\_(ツ)_/¯

[VerticalDiscussions]

If they are coming to fight the GTX 1080, unless they are gonna sell it for 1060 price, its gonna fall really deep into the deep and be left in there lol.

[TechGod]

6 minutes ago, Misanthrope said:

Oh but wait for:

 

1) DX12

2) drivers 

3) No not THAT DX12 game others like AOTS and....wait for the API to mature and devs to git gud! 

4) No not those drivers. Not those either...it takes like a year for them to mature! 

 

Ah the joys of AMD future proofing. 

Is AMD bashing fun? Do I get free stuff if I join in?

[vanished]

2 minutes ago, TechGod said:

This could be small Vega. 

 

#JustSayin

We can only hope  That would mean the big one would be even better, which is always good 

[TechGod]

Just now, Ryan_Vickers said:

We can only hope  That would mean the big one would be even better, which is always good 

I hope so too because otherwise Vega will fail if the big chip can't compete with the Ti

[DocSwag]

54 minutes ago, Droidbot said:

THAT'S NOT A POINT OF COMPARISON, DUMMY

 

AMD FineWineTM Technology will come in clutch and push V E G A over the line

I honestly hate FineWine Technology. It'd be better to just have those optimizations at the beginning like Nvidia than slowly roll them out over time. 

[Bananasplit_00]

29 minutes ago, Mr.Meerkat said:

Why not eh? Not like IPC matter anyway  

/s 

and thats why a Sempron 145 @ 7GHz  is the best CPU for games! who needs IPC when you have the gigely hertz! /s

[TheRandomness]

15 minutes ago, Misanthrope said:

Oh but wait for:

 

1) DX12

2) drivers 

3) No not THAT DX12 game others like AOTS and....wait for the API to mature and devs to git gud! 

4) No not those drivers. Not those either...it takes like a year for them to mature! 

 

Ah the joys of AMD future proofing. 

Though AMD is apparently working on good day 1 drivers... We can only hope.