|
Josh Lyman posted:Is v-cache faster than normal or is it just a way to get a ton of L3 cache? the latter
|
#
?
Mar 8, 2023 00:33
|
|
|
|
| # ? May 30, 2024 11:05 |
|
|
Combat Pretzel posted:Nope, haven't seen anything. If the BIOS has an ECC option to set, actual support for it seems to be part of AGESA 1.0.0.5C. Josh Lyman posted:Is v-cache faster than normal or is it just a way to get a ton of L3 cache? That said, it also consumes a not-insignificant amount of the packages thermal allowance (can't call it TDP because that term is rendered meaningless nowadays by companies), so the end-result is that the actual boost-clock is slower (since that's determined by how much thermal overhead each core and the complete package has). It comes down to whether you have a workload that's constrained purely by CPU cache, or something else.
|
|
#
?
Mar 8, 2023 00:58
|
|
|
what do you think about ZFS and v-cache? I know ZFS loves having more metadata and ARC available and I could see it lowering average latency on SSDs/etc due to immediate availability of metadata. I know ZFS kind of has a problem with that due to being optimized in the days of spinners and maybe a big ol cache would help. I can't find any benchmarks though. also what do you think about optane as a metadata-only L2ARC to extend that further?
|
|
#
?
Mar 8, 2023 05:38
|
|
|
Paul MaudDib posted:what do you think about ZFS and v-cache? I know ZFS loves having more metadata and ARC available and I could see it lowering average latency on SSDs/etc due to immediate availability of metadata. I know ZFS kind of has a problem with that due to being optimized in the days of spinners and maybe a big ol cache would help. I can't find any benchmarks though. For ZFS, RAM is basically king for caching. The 3D-cache is much too small, and not persistent enough for ZFS' use. Optane is amazing for ZFS. I have a few I got for really cheap. SLOG with forced sync writes to them. I have slower SSDs for L2ARC. You can also use them as a special vdev for metadata.
|
|
#
?
Mar 8, 2023 05:58
|
|
|
Paul MaudDib posted:what do you think about ZFS and v-cache? I know ZFS loves having more metadata and ARC available and I could see it lowering average latency on SSDs/etc due to immediate availability of metadata. I know ZFS kind of has a problem with that due to being optimized in the days of spinners and maybe a big ol cache would help. I can't find any benchmarks though. The cacheline on x86 is 64 bytes - that's the maximum datasize that the CPU is capable of working with. If some work fits into that 64 bytes and gets manipulated a lot, it gets a huge benefit from being able to fit in the L3 - and if the L3 is bigger, it can fit more of those sets of 64 bytes. Nothing about ZFS' design lends itself to fitting well into that workload - either data is clean and lives in memory, or it's dirty and needs to be written to disk.
|
|
#
?
Mar 8, 2023 13:34
|
|
|
Trip report with 7950X3D. Total changeout in case, cooler, everything except for the GPU (RTX 3080) and the NVME drives. Clean Windows 11 install. Previously: Corsair 800D i9900k AsRock z390 Taichi ultimate Corsair H150i Pro AIO 32 gigabytes Corsair DDR4 16-18-18-36 3200mhz RAM Now: Fractal Torrent 7950X3D AsRock B650e Riptide Wifi Icegiant Prosiphon Elite 64 gigabytes G.Skill Trident Z5 36-36-36-96 6000mhz RAM My typical usecase is extremely modded Skyrim, Total Warhammer 3, and Stellaris. Current takeaways: 1.) Won't break 67C in Prime95. 2.) Uplift of ~30fps (?!?!?!?!) in Skyrim [Note - one of the mods allows for over 60 fps without breaking the physics, and another one allows for use of DLSS and RTGI] 3.) Uplift of ~20fps in Total Warhammer 3 benchmarks. 4.) Have not tested Stellaris yet. Honestly did not expect this dramatic of an uplift in anything but late-game Stellaris, which is the main reason I upgraded. Super pleasantly surprised.
|
|
#
?
Mar 9, 2023 15:31
|
|
|
Late game Stellaris please. Would be cool if you could let it run for a few years on each CCD. Thx for reporting mate o7
|
|
#
?
Mar 9, 2023 18:05
|
|
|
I have not tried to buy new electronics on launch in....idk ever? But I'd like to get a 7800x3D on launch, anyone have any recommendations on that? I do live near a microcenter if that is an option.
|
|
#
?
Mar 9, 2023 20:11
|
|
|
Camp out and carry weaponry to beat back the ravenous hordes of scalpers.
|
|
#
?
Mar 9, 2023 20:20
|
|
|
I can't imagine the 7800x3d is going to be that bad, is it? I'd imagine if you hit up Newegg you could probably snag one. In store stock is always a gamble.
|
|
#
?
Mar 9, 2023 20:24
|
|
|
AutismVaccine posted:Late game Stellaris please. Would be cool if you could let it run for a few years on each CCD. Sure, might be a bit tho - have to reinstall Irony and get my mod load order resorted, then actually play ti late game. Which will be more difficult because I am also at home with a 3 month old right now!
|
|
#
?
Mar 9, 2023 20:27
|
|
|
Usually my Microcenter has enough stock for the "desirable" parts to make it to midday or so. If you try to get there first thing you might be good, but admittedly I've never actually tried this myself.
|
|
#
?
Mar 9, 2023 21:18
|
|
|
Guess who's mad again.
|
|
#
?
Mar 9, 2023 21:24
|
|
|
Rakeris posted:I have not tried to buy new electronics on launch in....idk ever? But I'd like to get a 7800x3D on launch, anyone have any recommendations on that? I do live near a microcenter if that is an option. I got to my local Microcenter five minutes before it opened and was able to get a 7950x3D no problem.
|
|
#
?
Mar 9, 2023 21:59
|
|
|
Combat Pretzel posted:Guess who's mad again.
|
|
#
?
Mar 9, 2023 22:41
|
|
|
Combat Pretzel posted:Guess who's mad again. I have no idea what he gets out of it.
|
|
#
?
Mar 9, 2023 23:41
|
|
|
Rakeris posted:I have not tried to buy new electronics on launch in....idk ever? But I'd like to get a 7800x3D on launch, anyone have any recommendations on that? I do live near a microcenter if that is an option. the 7950x3D was available for about an hour on Newegg before selling out and was restocked multiple times throughout the day afterwards. The 7800x3D will no doubt be more popular but unless AMD severely underestimates by how much you probably won't have an issue. they usually put up new items a few minutes after 9 AM ET
|
|
#
?
Mar 10, 2023 00:09
|
|
|
Combat Pretzel posted:Guess who's mad again.  Die mad, userbenchmark. Dude has just become increasingly unhinged over the years.
|
|
#
?
Mar 10, 2023 00:49
|
|
|
HalloKitty posted:I have no idea what he gets out of it. IIRC back when the madness started the site was running a lot of intel ads / sponsorships, and I thought that had something to do with it. But these days they have nothing but generic google adsense spots. So at this point I'd guess it's mainly about feeding a persecution complex.
|
|
#
?
Mar 10, 2023 01:24
|
|
|
Lockback posted:I can't imagine the 7800x3d is going to be that bad, is it? I'd imagine if you hit up Newegg you could probably snag one. It's probably going to sell out the initial stock like the 7950X3D did. Just F5 Newegg at launch (9AM ET as pointed out above) and you'll be fine, though.
|
|
#
?
Mar 10, 2023 01:43
|
|
|
Not late game, but starting out heavily modded Stellaris (including some performance enhancing mods) goes about a month a second at highest time tick on my setup. Edit: That's on a 1000 star galaxy. Gyrotica fucked around with this message at 03:50 on Mar 10, 2023 |
|
#
?
Mar 10, 2023 03:22
|
|
|
The userbenchmark guy is almost certainly a true believer (by now, even if he wasn't before) you don't get that kind of dedication from a mercenary
|
|
#
?
Mar 10, 2023 03:35
|
|
|
i dedicate all my lifes loyalty and fervor to Toy Corporation #42 Toy Corporation #42 i would die for you
|
|
#
?
Mar 10, 2023 05:05
|
|
|
Gyrotica posted:Not late game, but starting out heavily modded Stellaris (including some performance enhancing mods) goes about a month a second at highest time tick on my setup. a month a second? do you mean day maybe, else it is pretty bonkers.
|
|
#
?
Mar 10, 2023 08:44
|
|
|
A month a second is in the ballpark of what I get in unmodded early-game stellaris on my 5600X. A day a second would be unplayably slow. This is Stellaris, not HoI4.
|
|
#
?
Mar 10, 2023 08:48
|
|
|
CaptainSarcastic posted:
I'm almost disappointed they don't have any Apple CPU entries.
|
|
#
?
Mar 10, 2023 16:19
|
|
|
I want to hear that guys criticism of the 96 core epyc vs Intel.
|
|
#
?
Mar 10, 2023 16:41
|
|
|
HalloKitty posted:I have no idea what he gets out of it. The voices quieten down for a bit.
|
|
#
?
Mar 10, 2023 17:05
|
|
|
Perplx posted:I want to hear that guys criticism of the 96 core epyc vs Intel.
|
|
#
?
Mar 10, 2023 18:04
|
|
|
There must be new stock in European shops. I just got a tracking number for that 7950X3D I had backordered.
|
|
#
?
Mar 11, 2023 16:18
|
|
|
gradenko_2000 posted:The userbenchmark guy is almost certainly a true believer (by now, even if he wasn't before) I don't know if true believer is even strong enough, that reads almost like untreated schizophrenia.
|
|
#
?
Mar 11, 2023 18:09
|
|
|
BlankSystemDaemon posted:Better yet, the 128-core Ampere Altra Max at 3GHz - both more cores, and faster than either AMD or Intel. "Overgrown phone processor, can't run the only OS that matters, Windows 7"
|
|
#
?
Mar 12, 2023 01:40
|
|
|
Khorne posted:The part that annoys me most is that a 7950x3d will work better than a 7800x3d or 7950x for my use case. However, it's a product that is only appealing because they didn't release the product that best fits my use case. I'm sitting out until zen5 due to it. I was listening to MLID's (yeah, I know) most recent podcast tonight and he had an xbox studio engineering lead on and one of the things that he said was that he personally would rather have the dual-cache as well, but also that he thinks he would have uses for both clock-optimized ("7950XT") and the cache-optimized ("7950X3D") variants even just within his own studio... compiling loves X3D but there's also other stuff that is just number crunching (probably CAD work/graphics design). He thinks the X3D is really a fence-sitting SKU that satisfies neither camp, he thinks most people have a strong preference and they know whether they want cache or clocks and most people don't want the SKU that sits in the middle. And most people would probably pay more for a fully-clock-optimized XT with two good CCDs or two cache CCDs. And my take is, that leaves the possibility of heterogeneous as a fence-sitting base tier that is OK at both but lower-cost than fully optimized either way. (the other interesting nugget was, his opinion is don't buy the 7900X3D, because effectively that's a pair of 6-core chips and his opinion is we're about to shift into next-gen titles hard and 6-core chips are going to "begin screaming" as he put it. I think that really applies equally to the 7900X non-3D, the CCD barrier is kind of a cliff and games don't like suddenly having big latency jumps moving between CCDs, it's the same stuff you're talking about with discord audio crackling. So like, 7900X is a little iffy to begin with but a premium-priced dual-6C chip is an idiot trap... and really I have not encountered a single person who thinks 7900X3D is a good SKU as currently imagined.) Obviously you can Process Lasso processes to whichever CCD you like, but it'd be nice if Ryzen Master did that automatically. Ryzen Master can "help" windows scheduler too, doesn't have to ride totally on Windows, yeah we all know Microsoft are fuckups on the scheduler and I don't see that changing. To clarify one thing I said: I'm not sure heterogeneous will go away forever, I think there's a good chance XT and (fully-)X3D SKUs will become the premium models because that's a way to juice a little more ASP out of the silicon they have, and there's clearly customers for both products. But I could see heterogeneous becoming the base-tier for 7900X and 7950X especially with a poo poo-tier bin on the cache die (although again, that becomes a separate binning stream) but there's also arguments against doing that, it's more expensive than the current X configuration (although maybe if you're cranking out+stacking tons of cache it becomes cheaper if it's basically the default) and still loses to the current X configuration. But paradoxically maybe making the base tier shittier actually makes the premium models look better, it's a decoy product, nobody has to actually buy it, it just funnels people to the actual SKUs. I guess it doesn't make a ton of sense but who knows, big cache is a game-changer for some stuff, other people would love the flamethrower 6 ghz all-core 7950XTXX, and it's a way to push ASPs at a time when companies are desperate for that, I think fully-optimized versions are going to be available soon regardless of what happens with heterogeneous. On that note: Phoronix did a "for the things it helps, how much does it help?" piece. This is obviously a bit of a problematic number to try and wrap your head around, because it's inherently cherrypicked, it's "how big a number once we drop everything that's below average". But basically for those tasks the 7950X3D is around 18.5% faster (geomean) than the non-X3D model. Paul MaudDib fucked around with this message at 09:16 on Mar 12, 2023 |
|
#
?
Mar 12, 2023 02:31
|
|
|
Paul MaudDib edited this out posted:(anyway to be clear I listen to MLID with a truckload of salt, and there's many times he or his guests say things I don't really agree with, they're just humans and nobody's always right. In particular I just don't take any numbers seriously at all unless he's doing analysis on public data and I agree with the napkin math approach he's doing. But I see some unpopular points I've made come up every now and then too. It's mostly interesting to see guests in various spaces and what they think about current events and tech direction based on their own fields of expertise, there are a few of these little nuggets even if I don't take it all that seriously on the numbers/etc. Having a server procurement guy on was another interesting one recently, he had wendell on a while ago, etc. Again just like his retail "sources" you also have to keep in mind whether they have any particular knowledge of an area or it's just opinion... does the server guy know anything about TSMC wafer pricing? Probably not. So IDK maybe not a great primary source but there's some bits there if you're otherwise already informed and willing to discard stuff that is/might be BS. Mostly just something to listen to, since he puts out a couple hours a week I can put it on and listen while I clean or whatever. The LTT and Level1techs and GN shows and similar are bad imo, I already know everything they are gonna say and they don't add that much to the articles as a cast imo. I do like tech tech potato/ian cutress but he's not an "hours every week" level producer.) 
|
|
#
?
Mar 12, 2023 08:57
|
|
|
So my 6x2 5900x is already bad for gaming huh.
|
|
#
?
Mar 12, 2023 09:18
|
|
|
Threadkiller Dog posted:So my 6x2 5900x is already bad for gaming huh. no, especially if you got a good deal on it or have been enjoying it for a lot of years. but in a value-oriented market a premium midrange SKU at $550 or whatever is a tough sell when you could go cheaper with a 7900X or do the premium thing right with a 7950X3D. and the lifespan will be shorter than someone paying full price for a 12C in 2023 probably would want, if you don't want to do 8C for gaming then do 16C. the dev's thesis there is basically that hardware requirements are about to break loose a bit again as they cut PS4/XB1 legacy titles loose. And he might not be wrong either, we're pretty far into this "next-gen", it won't be all that many years until PS6 and a new xbox, and there's like, a total of 4 next-gen exclusives so far or something. (why do they call it ps5? less than 5 exclusive titles  )ok a few more have announced now. Paul MaudDib fucked around with this message at 10:12 on Mar 12, 2023 |
|
#
?
Mar 12, 2023 09:26
|
|
|
Yeah makes sense, though still seems strange that the double ccx should make such a material difference unless youre looking at HF gaming. I mean i usually do frame limiting anyway for noise purposes. But yeah seems prudent to go for an 9800X3D or whatever next time around like everyone else.
|
|
#
?
Mar 12, 2023 09:32
|
|
|
As always, core count isn't what really matters. It's single-threaded performance for lightly threaded games and overall multithreaded performance for multithreaded games, and core count is just part of the math that gets you one of those other figures. So I'm definitely not buying the idea that the 7900X3D will be suboptimal for gaming in the future just because there are only six gaming cores. The 5600X can basically run any game at 4K without issue as long as you don't have a 4090 (I can personally attest to my 4090 running at ~80% in a few games), and the benchmarks we have available back this up. The much faster 7900X3D will be just fine going forward—it's basically 2.5 generations ahead of the console CPUs while lacking only one gaming-capable core compared to them. edit: and to be clear, the new chipset drivers essentially already do automatically process lasso the cores to one CCD when gaming on an X3D chip. Or they're supposed to. Sometimes this seems to break and some games don't run on the correct CCD, but that issue will hopefully get ironed out over time. Dr. Video Games 0031 fucked around with this message at 10:24 on Mar 12, 2023 |
|
#
?
Mar 12, 2023 09:42
|
|
|
BlankSystemDaemon posted:Yep, Stanley Pain got it in one. again maybe I'm missing it but why would having a device-managed subset of L1ARC being superfast not help? Or is the problem that with a LRU-style cache eviction mechanism (or something similar) that the fact that metadata and actual data both live in L1ARC? But if there's actual metadata cachelines wouldn't those still be accessed frequently compared to actual data lines and thus stay in L3 better and benefit from a bigger cache? like i'm viewing it from a "the bigger the cache, the less you have to stop the world and go out to memory or beyond" and ZFS seems like it could benefit from stopping the world less and holding more metadata close to the core where it can be accessed quickly. It's not that it needs to be directly managed by ZFS itself, it's that the working set of metadata lines is probably quite large and de facto that's your working set, ZFS has to walk the tree to find other blocks. It's inherently the (tree-)pointer-chasing style of algorithm that a big L3 is supposed to benefit, right? Isn’t it faster to have your next pointer you want to chase cached in? Paul MaudDib fucked around with this message at 10:49 on Mar 12, 2023 |
|
#
?
Mar 12, 2023 10:18
|
|
|
|
| # ? May 30, 2024 11:05 |
|
|
Paul MaudDib posted:again maybe I'm missing it but why would having a device-managed subset of L1ARC being superfast not help? Or is the problem that with a LRU-style cache eviction mechanism (or something similar) that the fact that metadata and actual data both live in L1ARC? But if there's actual metadata cachelines wouldn't those still be accessed frequently compared to actual data lines and thus stay in L3 better and benefit from a bigger cache? Also, for what it's worth, ARC is MFU+MRU whereas L2ARC is just FIFO - so I think you're mixing things up a bit?
|
|
#
?
Mar 12, 2023 12:32
|
|