If you play AAA games with kernel anti-cheat then Wine and Proton are basically non-starters. Cant sidestep that a vast majority of people want to play those games.
"Kernel anti-cheat" and "spyware" is exactly what i mean. Thry do have millions of players. But that's like 20-30 games out of 1000s that run just fine.
Honestly (apart from the general hardware parts pricing crisis) the problem is that you can get the 7800X3D for almost the same price or just slightly more. Especially if you have a Micro Center nearby you.
> Especially if you have a Micro Center nearby you
Do they have stores in every state? Even assuming they have, seems fair to assume maybe just 3-4% of people actually have a Micro Center close enough to visit, oddly specific tip :p
It’s standard practice on the internet when talking about building PCs to mention Micro Center. Either “it’s cheaper if you have a microcenter nearby” or “this same bundle is $50 cheaper at micro center”.
The 9800X3D iGPU is definitely handy in some situations, but I tried using it with a 4K wayland desktop for a little while and the performance is really not good enough just for that, unless you're OK with some sub-60hz redraw. I was hoping that the driver situation would be better than Nvidia (lots of issues failing to come back from sleep/unlock). The driver was more reliable but I couldn't live with the occasional choppy redraw. It's really not the kind of CPU that you'd buy and expect to use without a high-end GPU anyway.
It feels like the people doing the benchmarks are stuck in time. POV-Ray benchmarks but not AI? Most people today, including enthusiasts, don't even know what POV-Ray is.
It's a gaming chip, literally a lower-cost downgrade of an AI chip meant for gamers instead of AI.
Why would it benchmarked for AI in these circumstances?
Beside, the second page look like those unnecessary AI benchmark to me. I wouldn't recognise such for certain or be aible to estimate their quality, but it's what their summary say.
But as a gamer who hasn't been able to make a significant upgrade to my rig in 6 years due to memory price and redirected processor engineering funding (all of it due to AI), it really bother me.
I assume they meant that gaming is the main market segment it serves, which from my understanding is correct. The X3D chips have great performance in games, which are generally constrained by memory latency. It's, of course, still a general purpose CPU that can do many things.
It is indeed a desktop CPU chip, but the 3dV cache is real good for gaming, thus it's also marked in that region.
From the official AMD website:
> Whatever the setting or resolution, lead your team to victory with elite gaming performance from a powerful 8‑core AMD gaming processor featuring AMD 3D V‑Cache™ technology.
I don't know because I'm not a specialist in benchmarks. The article authors are supposedly specialists, so I'd expect them to identify appropriate benchmarks to make statements such as "while this is a good value for games and general purpose use, it fails for local AI as these benchmarks show", for example.
It’s a shame Windows is still so far ahead for game compatibility. Feels a shame to pair such a nice CPU with such a bloated OS.
"Far ahead"? DRM'ed spyware, yes.
Everything else? Not so much, thanks to the monumental effort of Wine and Proton projects.
If you play AAA games with kernel anti-cheat then Wine and Proton are basically non-starters. Cant sidestep that a vast majority of people want to play those games.
"Kernel anti-cheat" and "spyware" is exactly what i mean. Thry do have millions of players. But that's like 20-30 games out of 1000s that run just fine.
Already down to $280 https://www.reddit.com/r/buildapcsales/comments/1uz0grn/cpu_...
Honestly (apart from the general hardware parts pricing crisis) the problem is that you can get the 7800X3D for almost the same price or just slightly more. Especially if you have a Micro Center nearby you.
But it's not bad if you can find a good combo deal with a mobo and RAM, https://www.newegg.com/p/pl?N=4812%204019%204020%204021&Srch...
> Especially if you have a Micro Center nearby you
Do they have stores in every state? Even assuming they have, seems fair to assume maybe just 3-4% of people actually have a Micro Center close enough to visit, oddly specific tip :p
It’s standard practice on the internet when talking about building PCs to mention Micro Center. Either “it’s cheaper if you have a microcenter nearby” or “this same bundle is $50 cheaper at micro center”.
A shame the x3d cache parts lack integrated GPUs. The extra cache could be handy for inference with smaller models.
Newer models do have both the x3d cache and integrated GPU. For example AMD Ryzen™ 7 9800X3D
The 9800X3D iGPU is definitely handy in some situations, but I tried using it with a 4K wayland desktop for a little while and the performance is really not good enough just for that, unless you're OK with some sub-60hz redraw. I was hoping that the driver situation would be better than Nvidia (lots of issues failing to come back from sleep/unlock). The driver was more reliable but I couldn't live with the occasional choppy redraw. It's really not the kind of CPU that you'd buy and expect to use without a high-end GPU anyway.
Caches take die space. GPU takes die space.
People buying gaming CPUs tend to not really want iGPU taking away thermals, so it makes kinda sense?
X3D cache doesn't take die space in traditional sense: it is a separate silicon packaged by stacking.
I wonder what perf. this cpu can get in bit/ternary inference, and whether the 512AVX units(or whats it's called on AMD), get throttled.
I’ve heard that AMD chips generate a ton of heat with AVX512 operations, overclockers turn off the instruction set for more stability.
It feels like the people doing the benchmarks are stuck in time. POV-Ray benchmarks but not AI? Most people today, including enthusiasts, don't even know what POV-Ray is.
It's a gaming chip, literally a lower-cost downgrade of an AI chip meant for gamers instead of AI.
Why would it benchmarked for AI in these circumstances?
Beside, the second page look like those unnecessary AI benchmark to me. I wouldn't recognise such for certain or be aible to estimate their quality, but it's what their summary say.
Why would they benchmark it in POV-Ray if gaming is all that matters?
People who buy computers may have a primary use for them but it's useful to know performance in their secondary workloads.
Actually it would be very interesting to know how it performs for AI and whether it has a better value.
I understand where you are coming from.
But as a gamer who hasn't been able to make a significant upgrade to my rig in 6 years due to memory price and redirected processor engineering funding (all of it due to AI), it really bother me.
Weren’t the prices normal until about a year ago?
It's not a gaming chip, it's a desktop CPU chip.
Imagine thinking that all people use PCs for is games...
I assume they meant that gaming is the main market segment it serves, which from my understanding is correct. The X3D chips have great performance in games, which are generally constrained by memory latency. It's, of course, still a general purpose CPU that can do many things.
It is indeed a desktop CPU chip, but the 3dV cache is real good for gaming, thus it's also marked in that region.
From the official AMD website:
> Whatever the setting or resolution, lead your team to victory with elite gaming performance from a powerful 8‑core AMD gaming processor featuring AMD 3D V‑Cache™ technology.
True, I suppose it is also useful for video editing and other uses.
The article describe it as a gaming chip, though. Might be because of the site's target audience? I don't know them.
> Might be because of the site's target audience?
Yes, but also because of official AMD marketing:
https://www.amd.com/en/products/processors/desktops/ryzen/70...
> with elite gaming performance from a powerful 8‑core AMD gaming processor
I would also call it a gaming CPU, because it is what it excels at. ¯\(ツ)/¯
What kind of AI benchmarks would you like to see?
I don't know because I'm not a specialist in benchmarks. The article authors are supposedly specialists, so I'd expect them to identify appropriate benchmarks to make statements such as "while this is a good value for games and general purpose use, it fails for local AI as these benchmarks show", for example.