Final Radeon RX Vega 64 Synthetic Performance Leaked – Trades Blows With A GTX 1080 Depending On Clock Speed

Author Photo
Aug 10, 2017

The final benchmark performance of the Radeon RX Vega 64 has been leaked by Videocardz (no surprises there), a few days before the launch. For all intents and purposes the Radeon RX Vega 64 trades blows with the GTX 1080, beating it on occasion or lagging behind depending on the clockspeed in question. This would mean that the RX Vega 56 would indeed go head to head with the Geforce GTX 1070 and the report we heard about it being faster than the 1070 might actually be true (depending on the clocks!).

Overclocked Radeon RX Vega 64 faster than a Geforce GTX 1080 Gaming X in 3DMark synthetics

The benchmarks tested were 3DMark FireStrike, 3DMark FireStrike Extreme, 3DMark FireStrike Ultra and 3DMark TimeSpy. Videocardz was able to find quite a lot of entries on the 3DMark database with quite a lot of clock variation. Depending on what clock the Radeon RX Vega 64 is running, it was either just behind the GTX 1080 or just after.

nvidia-geforce-rtxRelated NVIDIA Gained Major Discrete GPU Market Share With GeForce 10 Series GPUs Over AMD Radeon RX Series in Q2 2018

The high clocked variants was able to out-perform the GTX 1080 Gaming X on all benchmarks except Time Spy. Since these are synthetics, you can only gain so much valuable information but this at-least proves that speaking on raw available horsepower that developers will have access to – The RX Vega 64 and RX Vega 56 are very much current generation cards. They will not however, be taking the performance crown from NVIDIA thanks to the GTX 1080 Ti.

I have taken the liberty to extract the minimum and maximum scores across the clock variations for the FireStrike Ultra and Time Spy benchmarks so you can get an idea of what to expect from any RX Vega 64 that you end up buying:

Radeon RX Vega 64 3DMark FireStrike Ultra

All benchmarks are courtesy of Videocardz.com:

vega-64-mining-moneroRelated Major AMD Radeon Add-in-Board Partner TUL Corporation Lost 60% of Revenue In Wake Of Crypto Slump

  • MSI GTX 1080 Ti Gaming X (1924 MHz) – 7160 Points
  • AMD Radeon RX Vega 64 (1750 MHz) – 6057 Points
  • MSI GTX 1080 Gaming X (1924 MHz) – 5431 Points
  • AMD Radeon RX Vega 64 (1630 MHz) – 5353 Points
  • GTX 1070 iGame (1797 MHz) – 4421 Points

As you can see, the RX Vega 64 manages to put up an exceptional show. It is easily on-par with the GTX 1080 on stock clocks and outstrips it when overclocked. We are seeing a differential of almost 700 points which is pretty impressive considering this is FireStrike Ultra we are dealing with and anything above 500 points is pretty significant.

Radeon RX Vega 64 3DMark TimeSpy

All benchmarks are courtesy of Videocardz.com:

  • MSI GTX 1080 Ti Gaming X (1924 MHz) – 10013 Points
  • MSI GTX 1080 Gaming X (1924 MHz) – 7676 Points
  • AMD Radeon RX Vega 64 (1750 MHz) – 7615 Points
  • AMD Radeon RX Vega 64 (1630 MHz) – 6969 Points
  • GTX 1070 iGame (1797 MHz) – 6130 Points

TimeSpy is sort of the new standard for DX12 synthetics and here the RX Vega 64 manages to fit in neatly just behind the GTX 1080. While the difference between a stock RX Vega 64 (lowest clocks) is at most of 600 points, the GTX 1070 lags behind the Vega 64 by almost 800 – 1400 points depending on the clock rate. It is crystal clear that the positioning of this card is against the GTX 1080 which would imply we can expect the RX Vega 56 to offer similar performance against the GTX 1070.

AMD’s high end RX Vega 64 graphics card will contain 64 Compute Units, which (assuming the same ratio of CU to SPs as the current iteration of GCN) translates to exactly 4096 Stream Processors. The internal codename of this GPU is GFX9 (in this case the device ID is 687F:C3). Remember all our internal nomenclature analysis? Well, it’s the same thing, only in a more appealing format. Hawaii was GFX7, Polaris is GFX8 and RX Vega is GFX9.

AMD Radeon RX Vega 56 leaked benchmarks

We have already seen benchmark scores of the RX Vega 56 leaked by TweakTown and here is a recap for those interested:

The benchmarks were run at 2560×1440 with the AMD Radeon RX Vega 56 easily beating NVIDIA’s GeForce GTX 1070 in Battlefield 1, DOOM, Civilization 6, and even Call of Duty: Infinite Warfare. Tweak Town’s source said that Battlefield 1 was run on Ultra settings, Civ 6 was on Ultra with 4x MSAA, DOOM was at Ultra with 8x TSAA enabled, and COD:IW was running on its High preset. The test rig had an Intel Core i7-7700K @ 4.2GHz coupled with 16GB of DDR4-3000MHz RAM.

  • Battlefield 1: 95.4FPS (GTX 1070: 72.2FPS)
  • Civilization 6: 85.1FPS (GTX 1070: 72.2FPS)
  • DOOM: 101.2FPS (GTX 1070: 84.6FPS)
  • COD:IW: 99.9FPS (GTX 1070: 92.1FPS)

The RX Vega 56, as the name suggests, has 56 CUs with a stream processor count of 3,584. The Radeon RX Vega 56 will also feature 8 GB of HBM2 VRAM which will deliver a rated bandwidth of 410 GB/s along a 2048-bit bus interface. It will  be using HBM2 just like its big brother and will retail for $$399. The chip will feature a total compute horsepower of 10.5 TFLOPs (FP32) and will feature a TDP of 210W which will be provided through an 8 and 6 pin power connector configuration.

This means that it is competitively priced to go against the GTX 1070 which originally retailed for $449. If these numbers are true than we are easily looking at one of the performance per $ leaders of this generation. Just like the R9 290 and the Fury, it looks like AMD has reserved the value sweet spot for the second most powerful GPU in its lineup.

RX Vega score comparison gallery

AMD Radeon RX Vega 64 and Vega 56 Graphics Card Lineup:

Graphics Card AMD Radeon R9 Fury X AMD Radeon RX Vega Nano AMD Radeon RX Vega 56 Reference AMD Radeon RX Vega 64 Reference AMD Radeon RX Vega 64 Limited AMD Radeon RX Vega 64 Liquid
GPU Fiji XT Vega 10 Vega 10 Vega 10 Vega 10 Vega 10
Process Node 28nm 14nm FinFET 14nm FinFET 14nm FinFET 14nm FinFET 14nm FinFET
Compute Units 64 TBD 56 64 64 64
Stream Processors 4096 TBD 3584 4096 4096 4096
Raster Operators 64 64 64 64 64 64
Texture Mapping Units 256 TBD 224 256 256 256
Clock Speed (Base) 1000 MHz TBD 1156 MHz 1247 MHz 1247 MHz 1406 MHz
Clock Speed (Max) 1050 MHz TBD 1471 MHz 1546 MHz 1546 MHz 1677 MHz
FP32 Compute 8.6 TFLOPs TBD 10.5 TFLOPs 12.6 TFLOPs 12.6 TFLOPs 13.7 TFLOPs
FP16 Compute 8.6 TFLOPs TBD 21.0 TFLOPs 25.2 TFLOPs 25.2 TFLOPs 27.4 TFLOPs
Memory (VRAM) 4 GB HBM1 8 GB HBM2 8 GB HBM2 8 GB HBM2 8 GB HBM2 8 GB HBM2
Memory Bus 4096 bit 2048 bit 2048 bit 2048 bit 2048 bit 2048 bit
Bandwidth 512 GB/s TBD 410 GB/s 484 GB/s 484 GB/s 484 GB/s
TDP 275W 150W 210W 295W 295W 350W
Price $649 TBD $399
($499 US Actual)
($599 US Actual)
$599 $699
Launch 2015 2018 2017 2017 2017 2017