@gamecubepad said:
4k gaming on the PS4 Pro has 150W total system power consumption. Same game is 150W on PS4. Both systems were $399 competing against $500 systems from MS.
So not only does the total system need to consume only around 150W, it also needs to fit within a ~$380 BoM. Next gen systems will have something like 8-core 3rd-gen Ryzen CPU, 16GB GDDR6, 2GB sideport DDR4, 2TB HDD, and UHD Blu-Ray. Whatever is leftover from that power consumption and cost will determine what GPU they can include.
These are rumors of course, same rumors that said no 7nm Vega for gamers, so that was wrong, but they give a decent idea of where a PS5 GPU could slot. P.S.-Given the 7nm Radeon VII, these power consumption and price points seem very hopeful.
RX 3080 | Navi 10 | 8GB GDDR6 | 150W | RTX 2070/ GTX 1080 | $249 |
RX 3070 | Navi 12 | 8GB GDDR6 | 120W | GTX 2060/GTX 1070 | $199 |
Polaris 10 was also "supposed" to be a 150W part, and they had to downclock it 25% to get it into the PS4 Pro, and on the X1X, they had to use Hovis method and vapor chamber cooler with 384-bit memory bus to get stock performance. I don't think $399 gets you Hovis and vapor chamber cooler. Also, RTX 2060 is a $350 card, so they won't be hitting that performance with a $199 card now that we've seen their 7nm Vega pricing.
From https://wccftech.com/exclusive-first-amd-navi-gpu-will-have-40-cus-and-is-codenamed-navi-12/
NAVI 12 has 40 CU with unknown chip size, but X1X GPU is already 44 CU with 280 mm2 on 16 nm process node.
Extra transistor budget with 7nm could go towards upgraded raster engines/ROPS, deep learning instruction set(e.g. ray trace de-noise), and BVH search tree accelerator (e.g. ray trace)
VII has other workstation hardware features which largely useless for consoles consoles like half rate FP64 and high bandwidth cache lower/slower/higher latency than L2 cache.
VII consumed extra 700 million transistors over Vega 64's 12.5 billion transistors for octa rate INT4 and half rate FP64.
Expensive part with VII is the four stack HBM v2. AMD could develop cheaper VII 56 at 1800Mhz with two stack HBM v2 512 GB/s 8GB model. It would be faster than Vega 56 at 1710 Mhz which beaten Strix Vega 64
For 7nm estimate, let's use AMD's conservative 70% shrink of 14nm Vega 10 (includes useless high bandwidth cache),
1. X1X's 44 CU GPU with 280 mm2 size yields 196 mm2, hence 64 CU version yields 285 mm2, hence it's 9.6 TFLOPS or 60 CU active yields 9.0 TFLOPS
Looking at 1.5X TFLOPS gain instead over X1X at the same clock speed.
X1X GPU has extra support for FP10.
2. Upgrade to 64 ROPS with 8 MB L2 cache would more than double effectiveness over X1X's 32 ROPS with 2MB render cache and TMU's 2MB L2 cache.
Looking at +2.0X ROPS gain over X1X at the same clock speed
AMD should have mastered 64 ROPS with 256 bit DDR6-16000 (352 GB/s) or 384 bit DDR6-12000 (576 GB/s). RTX 2060 uses 14000 rated GDD6.
With the same 384 bit bus as X1X's but with GDDR6-12000 substitution, yields 1.77X gain.
Log in to comment