Tuesday, June 11th 2024
Possible Specs of NVIDIA GeForce "Blackwell" GPU Lineup Leaked
Possible specifications of the various NVIDIA GeForce "Blackwell" gaming GPUs were leaked to the web by Kopite7kimi, a reliable source with NVIDIA leaks. These are specs of the maxed out silicon, NVIDIA will carve out several GeForce RTX 50-series SKUs based on these chips, which could end up with lower shader counts than those shown here. We've known from older reports that there will be five chips in all, the GB202 being the largest, followed by the GB203, the GB205, the GB206, and the GB207. There is a notable absence of a successor to the AD104, GA104, and TU104, because NVIDIA is trying a slightly different way to approach the performance segment with this generation.
The GB202 is the halo segment chip that will drive the possible RTX 5090 (RTX 4090 successor). This chip is endowed with 192 streaming multiprocessors (SM), or 96 texture processing clusters (TPCs). These 96 TPCs are spread across 12 graphics processing clusters (GPCs), which each have 8 of them. Assuming that "Blackwell" has the same 256 CUDA cores per TPC that the past several generations of NVIDIA gaming GPUs have had, we end up with a total CUDA core count of 24,576. Another interesting aspect about this mega-chip is memory. The GPU implements the next-generation GDDR7 memory, and uses a mammoth 512-bit memory bus. Assuming the 28 Gbps memory speed that was being rumored for NVIDIA's "Blackwell" generation, this chip has 1,792 GB/s of memory bandwidth on tap!
The GB203 is the next chip in the series, and poised to be a successor in name to the current AD103. It generationally reduces the shader counts, counting on the architecture and clock speeds to more than come through for performance; while retaining the 256-bit bus width of the AD103. The net result could be a significantly smaller GPU than the AD103, for better performance. The GB203 is endowed with 10,752 CUDA cores, spread across 84 SM (42 TPCs). The chip has 7 GPCs, each with 6 TPCs. The memory bus, as we mentioned, is 256-bit, and at a memory speed of 28 Gbps, would yield 896 GB/s of bandwidth.
The GB205 will power the lower half of the performance segment in the GeForce "Blackwell" generation. This chip has a rather surprising CUDA core count of just 6,400, spread across 50 SM, which are arranged in 5 GPCs of 5 TPCs, each. The memory bus width is 192-bit. For 28 Gbps, this would result in 672 GB/s of memory bandwidth.
The GB206 drives the mid-range of the series. This chip gets very close to matching the CUDA core count of the GB205, with 6,144 of them. These are spread across 36 SM (18 TPCs). The 18 TPCs span 3 GPCs of 6 TPCs, each. The key differentiator between the GB205 and GB206 is memory bus width, which is narrowed to 128-bit for the GB206. With the same 28 Gbps memory speed being used here, such a chip would end up with 448 GB/s of memory bandwidth.
At the entry level, there is the GB207, a significantly smaller chip with just 2,560 CUDA cores, across 10 SM, spanning two GPCs of 5 TPCs, each. The memory bus width is unchanged at 128-bit, but the memory type used is the older generation GDDR6. Assuming NVIDIA uses 18 Gbps memory speeds, it ends up with 288 GB/s on tap.
NVIDIA is expected to double down on large on-die caches on all its chips, to cushion the memory sub-systems. We expect there to be several other innovations in the areas of ray tracing performance, AI acceleration, and certain other features exclusive to the architecture. The company is expected to debut the series some time in Q4-2024.
Source:
kopite7kimi (Twitter)
The GB202 is the halo segment chip that will drive the possible RTX 5090 (RTX 4090 successor). This chip is endowed with 192 streaming multiprocessors (SM), or 96 texture processing clusters (TPCs). These 96 TPCs are spread across 12 graphics processing clusters (GPCs), which each have 8 of them. Assuming that "Blackwell" has the same 256 CUDA cores per TPC that the past several generations of NVIDIA gaming GPUs have had, we end up with a total CUDA core count of 24,576. Another interesting aspect about this mega-chip is memory. The GPU implements the next-generation GDDR7 memory, and uses a mammoth 512-bit memory bus. Assuming the 28 Gbps memory speed that was being rumored for NVIDIA's "Blackwell" generation, this chip has 1,792 GB/s of memory bandwidth on tap!
The GB205 will power the lower half of the performance segment in the GeForce "Blackwell" generation. This chip has a rather surprising CUDA core count of just 6,400, spread across 50 SM, which are arranged in 5 GPCs of 5 TPCs, each. The memory bus width is 192-bit. For 28 Gbps, this would result in 672 GB/s of memory bandwidth.
The GB206 drives the mid-range of the series. This chip gets very close to matching the CUDA core count of the GB205, with 6,144 of them. These are spread across 36 SM (18 TPCs). The 18 TPCs span 3 GPCs of 6 TPCs, each. The key differentiator between the GB205 and GB206 is memory bus width, which is narrowed to 128-bit for the GB206. With the same 28 Gbps memory speed being used here, such a chip would end up with 448 GB/s of memory bandwidth.
At the entry level, there is the GB207, a significantly smaller chip with just 2,560 CUDA cores, across 10 SM, spanning two GPCs of 5 TPCs, each. The memory bus width is unchanged at 128-bit, but the memory type used is the older generation GDDR6. Assuming NVIDIA uses 18 Gbps memory speeds, it ends up with 288 GB/s on tap.
NVIDIA is expected to double down on large on-die caches on all its chips, to cushion the memory sub-systems. We expect there to be several other innovations in the areas of ray tracing performance, AI acceleration, and certain other features exclusive to the architecture. The company is expected to debut the series some time in Q4-2024.
141 Comments on Possible Specs of NVIDIA GeForce "Blackwell" GPU Lineup Leaked
But unless RTX 5080 uses a cut down version of GB202, it dosent seems RTX 5080 will be that beastly. That is of cause based on that it will be made with the GB203 gpu.
With cut down GB202 gpu i see however a potentiel beastly gpu that can or will beat rtx 4090. With GB203 i am not so sure. That will depend on IPC, clock speed and memory clock as well.
We can only guess. RTX 5090 looks to be a beast, but if other rumors are true, rtx 5090 can end up more exspensive than RTX 4090 as well.
Well no matter what, i dont exspect to upgrade to blackwell. It´s going to be a hell of an exspensive upgrade for me. Not only just for GPU, cause i need at least RTX 5080 for any meanful upgrade over RTX 4090. But also because i then would need a new CPU, do to my 5950X will defently cpu bottleneck a high-end blackwell gpu. So besides new gpu, i would need new cpu, motherboard, cpu cooler and memory. So no i think i shall just keep my curent setup with 5950X and RTX 4090 and be happy. Besides this combo is not so bad either yet.
Im afraid 5090 will indeed be more exspensive. 2000 USD or around there is my guess.Yeah i agree, that is also around there i would exspect the price to be. Specially if amd next gen is no match for Blackwell.
3080 Ti has been great, but it's time for an upgrade.
Hoping for around ~40% better performance than Ada, more is great of course.
3080 Ti to 5090/5080 Ti ideally around 2x faster.
Since I framecap to 237 FPS, faster/more efficient cards also means a lower total wattage which is always nice, unless new games/software push the GPU that much harder, which I doubt. Ada was a significant efficiency leap and very tempting, but I don't upgrade every gen of GPU.
"The meow you buy, the meow you save."
I don't know what you're looking at, but I'm seeing a slight uptick per tier, with all things increased except the bus width. GDDR7 makes up for part of the deficit though so bandwidth won't be worse than Ada relatively, that's good. But capacity, still 12GB in the midrange and 8GB bottom end? You're saying this is a good thing, now? Ada is already bandwidth constrained at the lower tiers. Nvidia is trying real hard to keep those tiers to what, 1080p gaming?
To each their own, but I think in 2025 people would like to move on from 1080p. The 8GB tier is by then bottomline useless and relies mostly on cache; the 12GB tier can't ever become a real performance tier midrange for long, its worse than the position Ada's 12GBs are in today in terms of longevity. Sure, they'll be fine today and on release. But they're useless by or around 2026, much like the current crop of Ada 12GBs.
As for AMD's inability to compete... RT fools & money were parted here. AMD keeps pace just fine in actual gaming and raster perf and is/has been on many occasions cheaper. They compete better than they have done in the past. Customers just buy Nvidia, and if that makes them feel 'screwed over'... yeah... a token of the snowflake generation, that also doesn't vote and then wonders why the world's going to shit.
You can't fix stupidity. Apparently people love to watch in apathy as things escalate into dystopia, spending money as they go and selling off their autonomy one purchase and subscription at a time.
Settle for nothing less! :cool:
IMO, there's no reason for these three variants to exist at all. I am 100% certain Nvidia will screw everyone on VRAM anyway, as part of the upsell:
The 5090 is 50% more compute than a 4090, so it'll cost $3000. Nvidia will probably say it's 60% faster which is why they can justify the asking price, which is irrelevant because China will buy them all for AI anyway and demand will outstrip demand until something better than 5090s comes along for AI datacenter racks.
I have no upgrade plans left for this year but sometime next year I wouldn't mind to upgrade my GPU and thats the highest I'm willing to go/what my budget allows. 'those will be plenty expensive enough where I live even second hand..:shadedshu:'
from 8gb vram at entry level but this probably is the going market size in terms of cost, 10 or at least 12 would be a fairer game.
The 5060ti or 5070 will probably cover my gaming needs for years to come, that is why i skipped the e.g. the 4060 ti.definitely curious to those two cards..