A number of details on the Radeon RX Vega have surfaced today, the Linux graphics driver reveales pretty precise specifications on the pending Radeon RX Vega GPU.
From the looks of it all, AMD Radeon RX Vega has 4 Shader Engines, 64 NCUs, 4 Render Back-Ends & 256 Texture Units. The funky g33ks from ComputerBase spotted this one after diving into the driver. If you check the code, this is the result set for Vega10 as spotted in the Linux drivers:
case CHIP_VEGA10:adev->gfx.config.max_shader_engines = 4;adev->gfx.config.max_tile_pipes = 8;adev->gfx.config.max_cu_per_sh = 16;adev->gfx.config.max_sh_per_se = 1;adev->gfx.config.max_backends_per_se = 4;adev->gfx.config.max_texture_channel_caches = 16;adev->gfx.config.max_gprs = 256;adev->gfx.config.max_gs_threads = 32;adev->gfx.config.max_hw_contexts = 8;
These specs and details are on par with what we have been telling for a long time now. 64 CUs x 64 shader units = 4096 shader processors. These are divided over four blocks. There is a total of 64 ROP units (16 per block) and the GPu is to get 256 Texture memory units. If you compare it, that pretty much twice a Radeon RX 570. The architecture details actually also show a lot of similarities with Fiji (Radeon R9 Fury (X)). Vega 10 is expected to battle the GeForce GTX 1070 and 1080.
 |
||||||
| AMD Radeon | RX Vega | RX 580 | RX 570 | RX 560 | RX550 | R9 390X |
| Graphics Core | Vega 10 | Polaris 20 XTX | Polaris 20 Pro | Polaris | Polaris | Grenada |
| Fabrication Process | 14 nm FinFET | 14 nm FinFET | 14 nm FinFET | 14 nm FinFET | 14 nm FinFET | 28 nm |
| Single Precision | 12.5 TFLOPs | 6.1 TFLOPs | 5.1 TFLOPs | ~2 TFLOPs | 3.48 TFLOPS | 5.9 TFLOPS |
| Memory | 8 GB HBM2 | 4/8 GB GDDR5 | 4/8 GB GDDR5 | 2/4 GB GDDR5 | 2/4 GB GDDR5 | 8 GB GDDR5 |
| Clock Frequency Base / Boost |
~1.5 GHz | 1,257 MHz 1,340 MHz |
1,168 MHz 1,244 MHz |
1,175 MHz 1,275 |
1183 MHz | 1,050 MHz |
| Compute Units | 64 | 36 | 32 | 14 | 8 | 44 |
| Shader Processors | 4096 | 2304 | 2048 | 1024 | 512 | 2816 |
| Memory Interface | 1024-bit per stack 8GB = 2 stacks |
256-bit | 256-bit | 128-bit | 128-bit | 512-bit |
| ROPs | 64 | 32 | 32 | 16 | 8 | 64 |
| TMUs | 256 | 144 | 128 | 56 | 32 | 176 |
| Memory Speed Effective | TBA | 8 Gbps | 7 Gbps | 7 Gbps | TBA | 6.0 Gbps |
| Memory Bandwidth | up-to 512 GB/s | 256 GB/s | 211 GB/s | 112 GB/s | TBA | 384 GB/s |
| TDP | 250 W | 185 W | 150 W | 75 W | TBA | 275 W |
| MSRP | TBA | $199 (4 GB) $229 (8 GB) |
$149 (4 GB) $179 (8 GB) |
$99 (2 GB) $119 (4 GB) |
$199 | $399 |
That chip is going to be huge though, roughly 500 mm² huge. To date it remains unclear when and where AMD RX Vega is to be released. Since shatter and noise is building up on the web, and with the AMD CEO mention that it is still good to go for a Q2 release, the Computex 2017 timeframe seems plausible for an announcement, we’re just not sure if that is going to be a hard launch though. At least it is good to see that the prediction for Vega10 have been spot on, with the 4096 shader processors and 8GB memory, really it should be good if they can keep the clocks high enough and the leakage low. Well that and the desired improved architecture efficiency. Computex 2017 starts late May.
