|
If AMD ever makes an APU die with 3D vcache on it, 100% of them be going in mobile gaming handhelds like the Ally. They see desktop APUs as the value products.
|
# ? Oct 21, 2023 04:18 |
|
|
# ? Jun 5, 2024 05:27 |
|
Does 3D cache significantly increase power draw?
|
# ? Oct 21, 2023 04:30 |
|
Yes but also no, because they have to run the cores under it at a lower voltage (and slightly lower clocks) to avoid thermal issues, so it evens out. I really do not think they'll ever do this to a low-power APU (or a desktop APU) anyway but if they wanted to burn some R&D money on a minuscule market for some reason, I think they probably could. The X3D chips they sell are ultra-efficient for what they do.
|
# ? Oct 21, 2023 04:44 |
|
kliras posted:threadripper announcement finally official I'm really tempted by the 24/48 because my setup needs stupid amounts of IO but not a trillion cores. I started off being annoyed that they cut 16 lanes out from the old threadripper before I re-read it and realized it's 48 PCIe 5 lanes with an additional 40 PCIe 4(?) They need to answer if these are arcwelded to the first MB they're installed in the way server chips are.
|
# ? Oct 21, 2023 04:49 |
|
BlankSystemDaemon posted:NUCA is a loving mess. What on earth are you complaining about here? It's the laws of physics which dictate that the costs, both energy and time, of accessing memory resources go up as the memories get bigger and/or further away. NUMA and NUCA aren't lazy designers deciding to make life difficult for everyone else, they're just consequences of this reality. The only way to achieve true uniformity in a large system is globally restricting performance in all cases just because you don't want CPUs to be capable of running faster while working with local (or relatively local) data.
|
# ? Oct 21, 2023 05:31 |
|
FuturePastNow posted:If AMD ever makes an APU die with 3D vcache on it, 100% of them be going in mobile gaming handhelds like the Ally. They see desktop APUs as the value products. Mmmmm.... disagree. Figure them to show up in Instinct MI APUs first, and then get binned down to the consumer market for something like Steam Deck 2.
|
# ? Oct 21, 2023 05:40 |
|
Would you even be able to use the 3dvcache as VRAM? Because if not I don't see the point - the IGPU is going to hold back the machine
|
# ? Oct 21, 2023 05:48 |
|
gradenko_2000 posted:Would you even be able to use the 3dvcache as VRAM? Because if not I don't see the point - the IGPU is going to hold back the machine Probably not. There's ways to work around the bandwidth bottleneck though - spinning the iGPU off to a separate chiplet and adding/stacking cache on the IO die, 7000 series style cache dies, stacked HBM (???). We'll see what AMD deems cost effective once Strix Halo is released.
|
# ? Oct 21, 2023 10:40 |
|
Adding even 2GB of HBM to the APU that's dedicated to the graphics would be pretty neat
|
# ? Oct 21, 2023 13:39 |
|
I wish mGPU had caught on beyond benchmarking utilities.
|
# ? Oct 21, 2023 16:23 |
|
Subjunctive posted:Jealous that they can fill all their DIMM slots, unlike my 7800X3D! have you tried again after the new microcode? supposedly memory is a lot better now. Kivi posted:Now if any of you US goons were willing to do a customs fraud and buy and send me a https://www.newegg.com/asrock-rack-romed8-2t/p/N82E16813140044 as that's also the lowest that board has been. oh gently caress, I knew about the mITX one but I didn't know romed8-2t was on sale... 😭 time to go look up epyc X and F sku pricing again... edit: they also have a dual-socket version on sale too. https://www.newegg.com/asrock-rack-rome2d16-2t-amd-epyc-7002-series-processors/p/N82E16813140060 BlankSystemDaemon posted:Sure, you can do cpusets/pinning/affinity, but then you're not using all those threads you bought the CPU to have access to, so what's the point? sure but the REAL question is whether you can do one socket with F skus and one socket with X, make your own big.little Paul MaudDib fucked around with this message at 20:24 on Oct 21, 2023 |
# ? Oct 21, 2023 20:21 |
|
Paul MaudDib posted:have you tried again after the new microcode? supposedly memory is a lot better now. no, last I looked people were having to set individual timings and try fifteen variants to get 6000 with 4x32, so I’ve been holding off I’ll find the other two DIMMs and try again when I get home next week
|
# ? Oct 21, 2023 20:32 |
|
Not sure where else to post this, but details on Qualcomm's upcoming Snapdragon X Elite have leaked: https://videocardz.com/newz/snapdragon-x-elite-leak-12-core-oryon-cpu-4-6tf-adreno-gpu-lpddr5x-av1-and-wifi7 My understanding is that this thing is meant to power Windows ARM laptops. Is the ARM branch of Windows no longer a joke? Won't most apps and games be wholly incompatible? I'm confused as to how this is supposed to be a good idea. What's the benefit of using an ARM processor over an x86 one?
|
# ? Oct 23, 2023 00:19 |
|
Windows 11 ARM is mostly used to run virtualized on Apple Silicon these days (and some Surface stuff, sure)
|
# ? Oct 23, 2023 00:26 |
|
windows ARM can run x86 binaries under emulation, but the performance is nowhere near as good as rosetta since non-apple hardware lacks the TSO extension that rosetta leans on if qualcomm does TSO and microsoft updates their emulator to use it then maybe they can get somewhere
|
# ? Oct 23, 2023 00:37 |
|
Dr. Video Games 0031 posted:Not sure where else to post this, but details on Qualcomm's upcoming Snapdragon X Elite have leaked: https://videocardz.com/newz/snapdragon-x-elite-leak-12-core-oryon-cpu-4-6tf-adreno-gpu-lpddr5x-av1-and-wifi7 Is there a native port of MS Office yet? I'd bet that for a whole lot of usage cases, if you've got an ARM browser and MS Office, then the substandard emulation for the long tail of other applications is fine. Come to think of it, anything that's .NET and JIT-ed anyway will be completely fine already too. I know that ARM is making inroads and gaining territory in the server segment, and clearly Apple has had enormous success. The last time that x86 and ARM went head to head was the Atom Phone days, but that was a weird market and time and Intel may well have sabotaged themselves. For Windows laptops, even if ARM doesn't gain market share quickly it'll limit AMD and Intel's profit margins.
|
# ? Oct 23, 2023 00:58 |
|
Iirc HP and Lenovo launched windows arm on snapdragon laptops in 2021. Battery life was very good for Windows (15hrs+) but you ran into the expected issues with emulation Seemed good enough for web browsing and Office work
|
# ? Oct 23, 2023 02:07 |
|
repiv posted:windows ARM can run x86 binaries under emulation, but the performance is nowhere near as good as rosetta since non-apple hardware lacks the TSO extension that rosetta leans on Yeah, I hope Qualcomm and Microsoft do since it not only benefits future customers buying Windows ARM machines, but also for those running Windows 11 ARM on ASi, like you said. I played around with a number of engineering apps I use under Windows 11 ARM emulation using my M1 Max; I had an issue with a few, but by and large the apps worked well and it was still pretty snappy.
|
# ? Oct 23, 2023 03:50 |
|
Canned Sunshine posted:Yeah, I hope Qualcomm and Microsoft do since it not only benefits future customers buying Windows ARM machines, but also for those running Windows 11 ARM on ASi, like you said. What software have you tried? I’ve been curious about SOLIDWORKS, but I haven’t found anyone who’s actually tried to use it outside a few old Reddit posts.
|
# ? Oct 23, 2023 04:04 |
|
I'm genuinely confused how a rumored Snapdragon relates to AMD. Pretty sure the A1100 ain't coming back. Did I miss something?
|
# ? Oct 23, 2023 05:32 |
|
mdxi posted:I'm genuinely confused how a rumored Snapdragon relates to AMD. Pretty sure the A1100 ain't coming back. Did I miss something? youre right, it should go in the Cyrix/IDT Megathread
|
# ? Oct 23, 2023 05:35 |
|
mdxi posted:I'm genuinely confused how a rumored Snapdragon relates to AMD. Pretty sure the A1100 ain't coming back. Did I miss something? I don't think we have an ARM thread?
|
# ? Oct 23, 2023 05:36 |
|
mdxi posted:I'm genuinely confused how a rumored Snapdragon relates to AMD. Pretty sure the A1100 ain't coming back. Did I miss something? Well, the leaked GPU performance on that SoC beats the RDNA3 on the Samsung Exynos 2400 in Vulkan performance. But this is the AMD CPU thread.
|
# ? Oct 23, 2023 05:43 |
|
gradenko_2000 posted:I don't think we have an ARM thread? https://forums.somethingawful.com/showthread.php?threadid=3970988
|
# ? Oct 23, 2023 05:56 |
|
I stand corrected, cheers
|
# ? Oct 23, 2023 05:56 |
|
gradenko_2000 posted:I stand corrected, cheers i also didnt know it existed, so you arent alone
|
# ? Oct 23, 2023 06:12 |
|
Cygni posted:i also didnt know it existed, so you arent alone It mostly popped up as a joke from a derail in either this or the Intel thread where someone decided we needed a VAX thread, so I doubled down.
|
# ? Oct 23, 2023 07:05 |
|
NewFatMike posted:What software have you tried? I’ve been curious about SOLIDWORKS, but I haven’t found anyone who’s actually tried to use it outside a few old Reddit posts. I've used CFD software (Ansys, Simflow) with varying levels of success, and then a few AutoDesk products (Revit, AutoCAD) that worked depending on the version, strangely enough (2022, 2024 worked for me, while 2023 had issues). I've used a few others too (WaterGems, Flow2D, HEC-RAS) without any real problems. I installed and ran Solidworks briefly using a prior model, but haven't really used it at all. It ran after I made a few adjustments I had seen online, but it did feel a bit slow at times with the model I used.
|
# ? Oct 23, 2023 21:05 |
|
mdxi posted:I'm genuinely confused how a rumored Snapdragon relates to AMD. Pretty sure the A1100 ain't coming back. Did I miss something? https://x.com/techmeme/status/1716536055351099605
|
# ? Oct 23, 2023 21:37 |
|
Kazinsal posted:It mostly popped up as a joke from a derail in either this or the Intel thread where someone decided we needed a VAX thread, so I doubled down. i don't know why we need two or more relatively slow CPU threads when the GPU thread lumps everything together and works fine, but IIRC there was pushback to a CPU megathread last time it came up
|
# ? Oct 23, 2023 21:46 |
|
ended up picking up this board for a build: https://pg.asrock.com/mb/AMD/B650E%20PG-ITX%20WiFi/ I assumed there would be a small air gap between the GPU backplate and the M2 fan assembly... but with the 3090 FTW3 Ultra, they are fully touching. Probably not great for my M2 drive that the 3090 also has the highest backplate temperatures in GPU history. That M2 cooling tower has become an M2 warmer... but the 3090 will love it! (theres lots of work arounds to this that ill probably do, just thought it was funny)
|
# ? Oct 23, 2023 23:19 |
|
Neither company needs to make new silicon to have something that can run windows 10
|
# ? Oct 23, 2023 23:58 |
|
hobbesmaster posted:Neither company needs to make new silicon to have something that can run windows 10 They want to get onboard for when everyone decides to switch to ARM so they can lock down platforms under the guise of a new architecture
|
# ? Oct 24, 2023 01:21 |
|
repiv posted:i don't know why we need two or more relatively slow CPU threads when the GPU thread lumps everything together and works fine, but IIRC there was pushback to a CPU megathread last time it came up The GPU factions co-exist better than the CPU ones, I think because AMD was not in recent memory (7800?) a winner, Intel was and therefore their partisans are more bitter about losing.
|
# ? Oct 24, 2023 01:26 |
|
case in point we currently have the same discussion about ARM happening concurrently in the AMD and intel threads
|
# ? Oct 24, 2023 01:32 |
|
It’s OK. We have 4 Ukraine threads, we can have this war in multiple threads too.
|
# ? Oct 24, 2023 02:13 |
|
Canned Sunshine posted:I've used CFD software (Ansys, Simflow) with varying levels of success, and then a few AutoDesk products (Revit, AutoCAD) that worked depending on the version, strangely enough (2022, 2024 worked for me, while 2023 had issues). I've used a few others too (WaterGems, Flow2D, HEC-RAS) without any real problems. Neat, thanks! banging fists on table NVIDIA CPU THREAD, NVIDIA CPU THREAD
|
# ? Oct 24, 2023 05:02 |
|
repiv posted:i don't know why we need two or more relatively slow CPU threads when the GPU thread lumps everything together and works fine, but IIRC there was pushback to a CPU megathread last time it came up The CPU tech & news stuff could easily be a megathread. The platfrom stuff -- "Is my CPU too hot? Overclock or undervolt? How do I run 4 sticks of ram at high speed?" -- is pretty domain-specific and can really get into the weeds. GPUs on the other hand pretty much run themselves, and there's TBQH very little difference between brand. They're either working or hosed, and when something is hosed there's not much you can do besides wait for new drivers / RMA the card.
|
# ? Oct 24, 2023 14:24 |
|
Cross linking from the Cities Skylines thread, my CPU scaling testing in Cities Skylines 2 with a 3090 at 1080p highish: Those 7800X3D results are painful, lol. Getting matched by a 1700. Game is laughably GPU bottlenecked.
|
# ? Oct 24, 2023 21:46 |
|
|
# ? Jun 5, 2024 05:27 |
Why is a sim game GPU bottlenecked? I play DF so I know I might not have much of a leg to stand on here, but surely a sim game doesn't need that amount of visual fidelity.
|
|
# ? Oct 25, 2023 01:51 |