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
Seems most people have 1080p or 1440p...
From TPU's latest 4070S review.
Alan Wake 2 RT @ 4K is what I would call an edge case.
If you want to play native 4K RT, you'd best be buying a 4070 Ti S or better lol.
Steam hardware survey had the GTX 1650 as the most popular GPU of 2023, and it was finally overtaken this year by the 3060, a 3-year-old GPU you can buy for $260, which is cheaper than even the cheapest 40-series offering.
AMD APUs and Intel IGPs have a far higher marketshare than the 4070Ti, 4080, 4090, and all three of those high-end 40-series GPUs combined are still outnumbered by 2016's GTX 1060.One thing to remember is that 4K monitors start at under $185 with good quality, reviewer-recommended models starting from about $250, and a monitor purchase can easily last a decade.
Someone can get themselves a decent 4K monitor and the cost increase over a half decent 1080p model may be under $100, which is only an extra $10 a year.
Only pure gamers who need to spend every single penny of their budget on GPUs will compromise their monitor purchase to try and buy a higher-tier GPU. For anyone who also works, enjoys movies, TV shows, or just likes lots of desktop real-estate, 4K monitors are both useful and extremely affordable. Nobody is forcing you to game at native resolution and now with DLSS/FSR/XeSS you can enjoy the benefits of a crisp, native resolution UI whilst the game itself renders at a resolution your GPU can handle, whatever that may be....
The lack of VRAM on cards lower on the stack would be expected but unfortunate. If I had to give people a tip, it would be that graphics have never been a deciding factor in whether a game is good or not and frankly returns are ever diminishing. It'd be one thing if we got actual AI in games (which would take up significant memory) that wasn't dumb as a bricks but slightly better looking graphics isn't worth taking the poison pills Nvidia wants you to swallow with it's VRAM on cards below the xx90 and their pricing.
I'm not dropping over a grand on that crap and still buying a 4K monitor...
I agree that the VRAM on the lower models is not enough, and the GB203 specs are too low.
What's your point?
But I'm planning to skip it.
GB202 = 28GB
GB203 = 24GB
GB205 = 20GB
GB206 = 16GB
GB207 = 12GB
This is ok now.
It is, although, curious that some people consider AMD drivers' bad reputation to be due to a "myth" when the amount of complaints is visibly superior to NVIDIA users. I've switched to NVIDIA shortly after and had no such episode since. Such a stark difference might sound ridiculous, but yeah, not once did I have a significant issue after updating NVIDIA drivers.
This is my personal experience and I notice more people complain about AMD drivers than NVIDIA's. Do with that information as you will.Would that I could, As said, this was years ago back in the RTX 20 era, no pics now. The monitor didn't explode, but there was no image displayed anymore. It was a frustrating experience figuring out what was going on, but I eventually switched to iGPU and was able to rollback or uninstall the latest Radeon drivers, don't remember which one was the solution.
Don't feel bad, everyone has had some kind of annoyance from minor to major issues with a driver on Nivida or AMD's side over the their lifetime of using them. Anyone that say otherwise is lying because as much as some people would believe, not every driver will work with every hardware/software configuration flawlessly.
I've had issues with Nvidia drivers that:
- broke video playback (nothing would play back in videos, just a black screen)
- caused stuttering
- caused flickering of images
- failed to allow secondary and tertiary monitors from waking up after going to sleep (the image wouldn't come back on them, I had to power cycle the computer to get the image to come back)
- Caused all blacks (shadows) in games to have green flickering
- Caused instability with SLI configurations
- gave me random driver failures with crashes to desktop
I learned some time ago that if a driver works and I don't get issues playing games, I have zero reasons to update the driver. There are times I run a driver for a year or more before I have to update it because I'm doing a re-install of Windows or swapping GPUs for whatever reason or maybe a new game I'm playing just won't run well without updated drivers.16 GB VRAM = 8 x 2GB chips.
20 GB VRAM = 10 x 2GB chips.
24 GB VRAM = 12 x 2GB chips.
28 GB VRAM = 7 x 2 x 2GB via clamshell.
Edit: not 6x32Bit but 8x32Bit. I hope...
Dunno what term is correct, but the screen was turned on with a black screen. No image. Maybe "the display broke" is not accurate, but that's what I meant.