AMD is the king of enterprise AI inference and now it's officially confirmed:
AMD makes the best AI inference hardware in the world.
Ally@treasureh8nter
$AMD MI355X beats Nvidia Blackwell in SemiAnalysis InferenceMax 🔥🔥🔥 #AMD #AI
English
AMDGPU
3.6K posts
AMDGPU
@AMDGPU_
AMDGPU is a community created by and for fans/enthusiasts of AMD processors and technologies. https://t.co/alPxz0TGTM
Bergabung Ekim 2021
294 Mengikuti5K Pengikut
Jensen Huang doesn't need a new chip. He needs a new moat.
Nvidia's most ambitious move from GTC was NemoClaw --a free, open-source AI agent platform designed to keep every company dependent on Nvidia's computing power, even as the chip competition heats up
Nvidia is becoming an operating system. The market is still pricing it as a chipmaker.
w/ @jaswu_
English
After a decade of releasing agonizing Nvidia propaganda to millions of viewers
Blow back from Nvidia's foundry tone deaf shilling finally reached critical mass.
TheGameVerse@TheGameVerse
🚨 Digital Foundry’s Richard Leadbetter says the team received death threats over their DLSS 5 coverage “We’ve received death threats… this is completely unacceptable. No one should face threats for doing their job or sharing tech.”
English
DLSS5 will not nearly have this fidelity. This is an offline AI render giving the hardware all the time it needs to complete.
DLSS5 is a simplified AI filter that has to run in realtime. It wil be much worse and have artifacts/glitching in fast scenes.
can@marmaduke091
🚨 DLSS 5 will look much better than this! Future of neural rendering will let devs, modders and player configure anything in the game This time, using an AI video model, reimagining GTA 4 in Russia, gives you a feel about how DLSS 5 might look like
English
@marmaduke091 DLSS5 will not nearly have this fidelity. This is an offline AI render giving the hardware all the time it needs to complete.
DLSS5 is a simplified AI filter that has to run in realtime. It wil be much worse and have artifacts/glitching in fast scenes.
English
AMDGPU me-retweet
🤪
Even a 2x performance boost is only about on par with AMD's CPUs from two years ago...
phoronix.com/review/nvidia-…
Semiconductor News by Dylan Martin@DylanOnChips
Jensen: "There are no CPUs in the world that are two times the performance of anything else, besides Vera." He doesn't see Nvidia's CPUs playing in the traditional server CPU market: "That's not the problem we're trying to solve."
English
AMDGPU me-retweet
Opinion: Nvidia DLSS 5 is a slap in the face to the art of video game design. bit.ly/4rA0WuA
English
AMDGPU me-retweet
@Grummz @3DCenter_org Welcome to a post consumerist tech industry brought to you by the stock market and the corruption it facilitates.
When stock market cap is mentioned more than actual revenue, share holders are now the customers.
English
I really find the way AI execs talk about AI at the expense of gaming extremely distasteful.
I like where the advances can take us but I hate the hubris, the lack of GPUs, the RAM prices, the absolute way Nvidia and the others just don't give a crap.
The money is insane and it makes them feel like they can do no wrong, when they really should be thinking about their customers more than their AIs.
I'm not anti-AI. I'm anti whatever this corporate attitude is.
English
@el_LEJAN Neural rendering is a general term with different meanings.
The term is not standardized across the industry. AMD neural rendering is not necessarily the same as Nvidia's.
This is irrelevant to the fact of AMDs clear focus on native gaming performance vs Nvidia AI reliance.
English
AMD continues to achieve breakthroughs in actual raw gaming performance per $
While Nvidia🤡 continues to spoof gamers with AI fakes (DLSS5)
TechPowerUp@TechPowerUp
AMD "RDNA 5" to Heavily Boost Shader Performance in Games with New Dual-Issue Pipeline tpu.me/zzfk
English
@NikTek @Sylvaneus The "neural arrays" is just a marketing name. They are just more powerful AI processors used for faster gaming features like texture compression, upscaling, frame gen and RT optimizations (redstone). they are also for LLM performance.
All in heavy demand from gamers, etc.
English
Jensen Huang has responded to the comments calling DLSS 5 AI slop:
"First of all, they're completely wrong. The reason for that is because, as I have explained very carefully, DLSS 5 fuses controllability of the geometry and textures and everything about the game with generative AI. It’s not post-processing, it’s not post-processing at the frame level, it’s generative control at the geometry level."
He has also added:
"This is very different than generative AI; it’s content-control generative AI. That’s why we call it neural rendering"
What are your thoughts on this?
English
@NikTek @Sylvaneus Lies.
AMD isn't doing the same thing.
They have never done this.
So many mouthbreathing gamers still don't get AMD as a company.
AMD only reacts in response to the market consensus (gamers). If there is big demand for a feature or capability it they will do it.
English
@TechPowerUp 🤣 Of course Jensen doesn't get it. Nvidia is not now nor have they ever been a real gaming company.
They are marketing con artists obsessed with peddling their toy chips at any cost.
English
What’s the best open source model for coding right now? Any size 😆
Matthew Berman@MatthewBerman
.@nvidia hand delivered a pre-production unit of the @Dell Pro Max with GB300 to my house. 100lbs beast with 750GB+ of unified memory to power the best open-source models in the world. What should I test first?
English
AMDGPU me-retweet
📈 GPU Retail Sales Week 9-11 ’26 (mf) 🇩🇪
GPU sales tank to a third of the usual volume.
ℹ️ Units
AMD: 1 110 units, 55.6%
Nvidia: 795 units, 39.8%
Intel: 90 units, 4.5%
ℹ️ by GPU
RX 9070 XT: 510 (25.6%)
RX 9060 XT: 405 (20.3%)
RTX 5080: 235 (11.8%)
RTX 5070 Ti: 180 (9.0%)
RTX 5070: 140 (7.0%)
RTX 5060: 140 (7.0%)
RTX 5090: 105 (5.3%)
RX 7600: 40 (2.0%)
RTX 3050: 60 (3.0%)
RX 7900 XTX: 30 (1.5%)
RX 7900 XT: 20 (1.0%)
RX 7700 XT: 20 (1.0%)
ARC B580: 20 (1.0%)
ARC B570: 20 (1.0%)
ARC A770: 20 (1.0%)
ARC A380: 10 (0.5%)
RX 6400: 10 (0.5%)
GT 1030: 20 (1.0%)
GT 710: 20 (1.0%)
GT 730: 10 (0.5%)
RTX 5050: 15 (0.8%)
ℹ️ VRAM Share (8 GB vs 16 GB+)
16 GB+: 1 510 units, 75.7%
≤ 8 GB: 485, 24.3%
ℹ️ Architecture Share
RDNA 4: 915 units, 45.9%
Blackwell: 815, 40.9%
RDNA 3: 140, 7.0%
Intel Arc: 90, 4.5%
Legacy (GT / low-end): 35, 1.8%
ℹ️ Board Partner Share
PowerColor: 370 units, 18.5%
MSI: 335, 16.8%
XFX: 310, 15.5%
Gigabyte: 255, 12.8%
Sapphire: 240, 12.0%
Palit: 215, 10.8%
ASRock: 180, 9.0%
Sparkle: 90, 4.5%
ASUS: 60, 3.0%
ZOTAC: 30, 1.5%
INNO3D: 30, 1.5%
ℹ️ Top 3 best selling SKUs
PowerColor Radeon RX 9070 XT Reaper 16 GB: 110 units
PowerColor Radeon RX 9060 XT Reaper 16 GB: 80 units
Sapphire Radeon RX 9070 XT Pulse 16 GB: 70 units
$AMD $INTC $NVDA
English
InferenceX, formerly InferenceMAX, is changing the industry
Performance of hardware + software is constantly evolving with new models and new ML Systems techniques
All major GPUs from AMD and Nvidia + soon TPUs / Trainium
AMD said they'd send a pic of Lisa with their belt soon!
SemiAnalysis@SemiAnalysis_
InferenceX v2: NVIDIA Blackwell Vs AMD vs Hopper - Formerly InferenceMAX, GB300 NVL72, MI355X, B200, H100, Disaggregated Serving, Wide Expert Parallelism, Large Mixture of Experts, SGLang, vLLM, TRTLLM semianalysis.substack.com/p/inferencex-v…
English
$1,000 for whoever comes up with the best name replacement for InferenceMAX
InferenceMAX 2.0 dropping soon but we have to rename it because HBO MAX sent us a cease and desist.
We have all NVIDIA GPUs from h100 to GB300 on large MoEs with SOTA optimizations like Disagg PD tested
English
AMDGPU me-retweet
$NVDA sells chips to a datacenter startup who pays by issuing junk bonds, and then Nvidia signs a contract to rent them back for the next 16 years even though they only last 4-6 years. What's the worst thing that could happen? 👀
English