|
It's depressing to realize that if you want a good motherboard you pretty much gotta give up and give Asus like 300 loving bucks  I mean, some of Asrock's high end stuff is also good, at least as far as power delivery goes, but the BIOSes really do seem worse over the board. Asus' midrange and low end stuff is pretty lovely and overpriced to boot, too.
|
#
?
Jul 26, 2018 20:20
|
|
|
|
| # ? Apr 28, 2024 03:58 |
|
|
It's buggy af. Neither of the 2 non alpha bios versions even report mem voltage to the system, the system is meant to reboot into safe mode if unstable settings fail post (doesn't work), even vcore offset doesn't work. The alpha has its own problems, setting memory voltage doesn't work  Wtf are they even doing. None of the bioses are release ready at all. GRINDCORE MEGGIDO fucked around with this message at 20:35 on Jul 26, 2018 |
|
#
?
Jul 26, 2018 20:28
|
|
|
eh. There's going to be a new crop of boards coming with 7nm soon enough.
|
|
#
?
Jul 26, 2018 20:34
|
|
|
TheFluff posted:It's depressing to realize that if you want a good motherboard you pretty much gotta give up and give Asus like 300 loving bucks Even today's high-end stuff isn't that high-end by the standards of previous generations. Back in the X58 days, a primo motherboard would have 16-24 phases, and the absolute best you get nowadays is like 8 phases, or maybe "12 phase"/doubled 6-phase, even on something like X299 or Threadripper that can suck down hundreds of watts. Paul MaudDib fucked around with this message at 20:40 on Jul 26, 2018 |
|
#
?
Jul 26, 2018 20:35
|
|
|
GRINDCORE MEGGIDO posted:It's buggy af. Neither of the 2 non alpha bios versions even report mem voltage to the system, the system is meant to reboot into safe mode if unstable settings fail post (doesn't work), even vcore offset doesn't work. Taking bribe money from Intel  This is a joke, but I'm sure on AMD Reddit this would be taken as fact without question
|
|
#
?
Jul 26, 2018 20:36
|
|
|
It's concerning me tbh. AMD seems to be executing well - but if they don't get good bios support, it's going to piss people off. I see other people saying similar on social media, with points like "replaced with an 8700k and everything works". It's not AMD at fault there tbh. I don't know how Asus is or how other premium asrock boards are, just that they really don't seem to care about the new board I just bought. GRINDCORE MEGGIDO fucked around with this message at 20:49 on Jul 26, 2018 |
|
#
?
Jul 26, 2018 20:39
|
|
|
GRINDCORE MEGGIDO posted:It's concerning me tbh. AMD seems to be executing well - but if they don't get good bios support, it's going to piss people off. Complete speculation: What might be happening is Intel giving a minimum level of quality that must be reached, and AMD is letting vendors do whatever they want, Intel might also subsidize support on boards that didn't sell well so they keep getting BIOS support (as long as they met their minimum requirements) AMD being in a bad position previously hasn't provided these guidelines yet. It also could be that Intel hasn't changed much in a decade and AMD is doing something new for once causing issues. Finally it's possible motherboard manufactures just don't have the resources allocated because they expected another bulldozer. I'm not sure how long it takes for a board to go from concept to stores but I'd imagine at least a year.
|
|
#
?
Jul 26, 2018 20:46
|
|
|
GRINDCORE MEGGIDO posted:It's concerning me tbh. AMD seems to be executing well - but if they don't get good bios support, it's going to piss people off. Buildzoid seemed to quite like the Asus Crosshair VII at the very least. He should, though, given that that's the highest end X470 board you can get.
|
|
#
?
Jul 26, 2018 21:01
|
|
|
There have definitely been some growing pains with X370 chipset BIOS code. The latest release and beta BIOSes for the ASRock X370 Taichi still have annoying bugs, particularly with overclocking settings (a primary selling point for the board), but I've found that overclocking works flawlessly via userspace tools i.e. Ryzen Master on Windows or ZenStates.py on Linux.
|
|
#
?
Jul 26, 2018 21:03
|
|
|
If it's anything to do with AMD / Intel working with board manufacturers, I hope AMD does more of that. I love this system, it's just frustrating to tally up the bios problems and think "you fuckers, this wasn't even cheap".
|
|
#
?
Jul 26, 2018 21:08
|
|
|
I think AMD is moving more towards having minimum standards or at least standard tiers of board design since IIRC, AMD is moving to have boards flashable even without a CPU.Paul MaudDib posted:AMD's earnings are in, and it's actually really good. Computing/graphics group revenue is down a bit and income is down about 15%, but enterprise/semicustom revenue increased by 26% and income more than quadrupled, which more than offset the C/G group. So what you're saying is that you were extremely hyperbolic, maybe even wrong, when stating AMD was overly exposed to crypto and EPYC wasn't being adopted fast enough?
|
|
#
?
Jul 26, 2018 21:54
|
|
|
its worth noting that the Z370 BIOS weren't great at launch either, even though it's a repackaged Z270
|
|
#
?
Jul 26, 2018 22:50
|
|
|
Cygni posted:its worth noting that the Z370 BIOS weren't great at launch either, even though it's a repackaged Z270 This board has been better then an Asus z170 itx. It's running 100% aok at stock.
|
|
#
?
Jul 27, 2018 00:57
|
|
|
FaustianQ posted:So what you're saying is that you were extremely hyperbolic, maybe even wrong, when stating AMD was overly exposed to crypto and EPYC wasn't being adopted fast enough? Woah woah, calling him wrong sounds like an ad hominem to me, please stick to the facts 
|
|
#
?
Jul 27, 2018 08:46
|
|
|
Paul MaudDib posted:You can only stack like 1/4 as many cores into a system as on the Intel platform (224C vs 64C). That's a pretty big deal in compute-intensive markets (HPC). Rack units matter. Eating more units on the interconnects matters. RAM capacity matters, on different problems than RAM bandwidth. Smaller CCXs have consequences.
|
|
#
?
Jul 28, 2018 17:06
|
|
|
pixaal posted:So it also cannot tell if the temp is negative, which is also a dangerous place to be. That seems really dumb. 125C is a really weird spot, I'd kind of expect it at 127/128 and using a signed byte. It's not using 1LSB per degree. The intel core temp is 1000LSB/degree, for example: Core 2: +29.0°C (high = +80.0°C, crit = +100.0°C) $ cat /sys/devices/platform/coretemp.0/hwmon/hwmon0/temp4_input 29000 Most of the sensor packages just dump the ADC value of the termistor and let you do the math in software. Going negative is probably a bios bug - the data comes in as signed/unsigned 12-bit in a 16-bit register, and if you don't put it into a 32-bit variable before doing math you'll overflow and wrap around. I run into that all the time in embedded - the implicit conversion rules are non-obvious.
|
|
#
?
Jul 28, 2018 17:58
|
|
|
Dafuq is this poo poo? Base clocks: 2950X 16C = 3.1GHz 2970X 24C = 3.5GHz 2990X 32C = 3.4GHz https://hwbot.org/hardware/processors#key=tr4 This gotta be wrong. Then again, they also claim 125W TDP on the 16C one.
|
|
#
?
Jul 30, 2018 00:51
|
|
|
EmpyreanFlux posted:So what you're saying is that you were extremely hyperbolic, maybe even wrong, when stating AMD was overly exposed to crypto and EPYC wasn't being adopted fast enough? I was right about C/G group revenue and profits being down. Enterprise computing was up, but it was a bunch of one-time revenue from a 2017 change in accounting practices (w/r/t when you can recognize revenue from clients). And to quote Devinder, the Enterprise group numbers won't be sustained into future quarters. quote:Devinder Kumar - Advanced Micro Devices, Inc. quote:Stacy Aaron Rasgon - Bernstein Research It's good that they could pull forward some revenue though! Probably a smart tactical move for them to pull forward some revenue to cover the losses they otherwise would have taken in Q2. Combat Pretzel posted:This gotta be wrong. Then again, they also claim 125W TDP on the 16C one. You're getting 50% more cores on the 2970X at the same TDP as the 1950X, at 100 MHz higher base-clocks. What are you complaining about? If that's how they actually measure out at those clocks, that's fantastic. Paul MaudDib fucked around with this message at 05:50 on Jul 30, 2018 |
|
#
?
Jul 30, 2018 05:37
|
|
|
https://www.youtube.com/watch?v=yiHscW5Scfc der8auer at it again, seems that TR2 will hit ~500W around the 4.0Ghz mark, all core. Should hit ~5700-5800 Cinebench, which is pretty much spot on for Zen+ IPC deficit and clock deficit from the 28C 1.5Kw Intel demo. Also the best configuration is to use 2Chs for 2 of the dies, and not 1Ch per die. Paul MaudDib posted:I was right about C/G group revenue and profits being down. Enterprise computing was up, but it was a bunch of one-time revenue from changing their accounting practices. And to quote Devinder, it won't be sustained into future quarters. Okay you're going to have to forgive me but that's some Fishmech level of weaslyness. Just because you were right about the vague direction of C/G and Enterprise doesn't make you anything close to an Oracle when you were off by an order of magnitude.
|
|
#
?
Jul 30, 2018 05:57
|
|
|
EmpyreanFlux posted:Okay you're going to have to forgive me but that's some Fishmech level of weaslyness. Just because you were right about the vague direction of C/G and Enterprise doesn't make you anything close to an Oracle when you were off by an order of magnitude. Lol, I mean, can you imagine what you'd be saying if the tables were turned and I was the one staking an argument on one-time balance-sheet push-backs?  indeed.I said C/G would probably lose about 15-30% (30-60% of the graphics half of C/G's revenue) and they lost 15% of C/G. Sorry I didn't know the intricacies of corporate finance law that they would be able to pull forward a bunch of one-time revenue (by AMD's own admission). TBH, it kinda seems like $AMD's perpetual opposite-day continues. Not the best quarterly return ever (on the fundamentals), not the best forecast, and they went up anyway. Maybe they're undervalued long-term but one-time revenue is probably not the way to show that. (I am certainly not claiming to have any special insight into financial poo poo though) Paul MaudDib fucked around with this message at 09:26 on Jul 30, 2018 |
|
#
?
Jul 30, 2018 08:28
|
|
|
I had a look to see if anyone had actually announced an Epyc decision for HPC. Last month the Italian National Institute for Nuclear Physics did so although without scale or timeline. https://insidehpc.com/2018/06/amd-epyc-powers-hpc-national-institute-nuclear-physics-italy/ They've gone for the AMD EPYC 7351 which is the two-socket 16 core variant. I reckon that'll be most common choice, as the best balance between the number of cores, all-core boost and memory bandwidth.
|
|
#
?
Jul 30, 2018 10:24
|
|
|
Paul MaudDib posted:Lol, I mean, can you imagine what you'd be saying if the tables were turned and I was the one staking an argument on one-time balance-sheet push-backs?  , I don't know what is weirder, you posting this then coming back an hour later to try and toss in a :sickburn: or that I'm making any claims at all or even would. If you're going to just slide in smugly into the AMD thread to consistently prognosticate about how doom and gloom and the cryptoapocalypse too™ are coming for AMD and then get proven wrong then you're going to get called out for it. But yeah, maybe I would hope people would be okay with calling me out on bullshit and not retreat into moving goal posts?
|
|
#
?
Jul 30, 2018 10:50
|
|
|
Paul MaudDib posted:You're getting 50% more cores on the 2970X at the same TDP as the 1950X, at 100 MHz higher base-clocks. What are you complaining about? If that's how they actually measure out at those clocks, that's fantastic. --edit: Someone figured out by messing with the support URLs, which Threadripper SKUs there's going to be. 2920X, 2950X, 2970WX and 2990WX. https://www.reddit.com/r/Amd/comments/933g2z/threadripper_2nd_gen_skus_2990wx_2970wx_2950x/ So I guess the base clocks and wattages I referred to earlier might not be 100% accurate, with what the site labeling the 2970WX and 2990WX wrong. Combat Pretzel fucked around with this message at 13:19 on Jul 30, 2018 |
|
#
?
Jul 30, 2018 12:23
|
|
|
I would note that "WX" was previously used as a prefix for AMD's professional GPUs. One could make the assumption that this was a code for "Workstation". This might imply that the 2970WX and 2990WX are being positioned differently than their predecessors.
|
|
#
?
Jul 30, 2018 15:55
|
|
|
|
|
#
?
Jul 30, 2018 19:34
|
|
|
"4.0Ghz" ".0912v" "Platinum Sample" I am making a joke, I know VCore fluctuates.
|
|
#
?
Jul 30, 2018 19:48
|
|
|
32cores HEDT, no 1000w chiller required
|
|
#
?
Jul 30, 2018 19:51
|
|
|
Crap, that is a fine processor.
|
|
#
?
Jul 30, 2018 20:03
|
|
|
rip my threads. rip em all.
|
|
#
?
Jul 30, 2018 20:08
|
|
|
In the comments, one of the Videocardz guys suggests that he's seen CPUs with 2990X silkscreened on them and that there's possibility two variants (well, X and WX). Would be loving hilarious, if there's going to be octochannel HEDTs.
|
|
#
?
Jul 30, 2018 20:35
|
|
|
Seamonster posted:rip my threads. rip em all. Obligatory RIP AND TEAR 
|
|
#
?
Jul 30, 2018 20:37
|
|
|
Thread count so high you could use it as a thrifty blanket. How do video games play with THREADRIPPER's? Is a normal processor of similar GHz better due to whatever reason involving the multichip set-up? I wish I had a reason to actually entertain the thought of getting a THREADRIPPER. But all I do is play video games, I don't even stream or do anything creative.
|
|
#
?
Jul 30, 2018 20:44
|
|
|
You absolutely don't need a Threadripper if all you do is straight gaming. A $190 Core i5-8400 is sufficient for that task.
|
|
#
?
Jul 30, 2018 20:48
|
|
|
Broose posted:Thread count so high you could use it as a thrifty blanket. It does nothing for games. Most games bottleneck hard on one thread. That is why the hex core Intel's get higher gaming benchmarks than the octo core AMDs in most games. For games the threads are asymmetrical. There is N number of minor threads doing small things and then usually one render or core engine thread that bottlenecks. So even though you have games that "use" 8 or 10 or even 20 threads you get more use out of single threaded performance than having a ton of cores.
|
|
#
?
Jul 30, 2018 20:51
|
|
|
Broose posted:Thread count so high you could use it as a thrifty blanket. Got any brothers? You could run 2 Windows VMs or more on the same system and run games that way. 
|
|
#
?
Jul 30, 2018 20:53
|
|
|
Mr Shiny Pants posted:Got any brothers? You could run 2 Windows VMs or more on the same system and run games that way. X399 mobos have 4 16x slots, can't you do a 4 seater all with pass-through video cards to the VM? I was going to actually price it out, but pcpartpicker won't let you put 4 video cards on one build.
|
|
#
?
Jul 30, 2018 21:17
|
|
|
Is there going to be a X499 chipset and related mainboards? It's been pretty silent about that. I suppose it'll continue on X399 except with better power delivery?
|
|
#
?
Jul 30, 2018 21:47
|
|
|
2000-series Threadripper doesn't have official ECC support, right? I wonder if the existence of X and WX variants of the same number indicates no-ECC/ECC support. I can't imagine *why* they'd be able to differentiate like that, though. Broose posted:Thread count so high you could use it as a thrifty blanket. If they do, that won't be for another console generation, at least. At least, that's the grand plan that's inferred on AMD's part, right? Now that Sony and Microsoft now both have consoles that use an x86 ISA, as long as they continue to control the console hardware, they can make studios architect their games and engines in certain ways to take best advantage of them. And then once games start going wide because that's the better way to do it on the consoles... but that's 1) not now, 2) entirely speculation and 3) entirely dependent on AMD regaining GPU market share. Single-threaded performance is still king, in most cases, and TR can't provide that like Intel can. Your rational mind is correct, you have no reason to get a Threadripper. SwissArmyDruid fucked around with this message at 13:49 on Jul 31, 2018 |
|
#
?
Jul 30, 2018 23:33
|
|
|
SwissArmyDruid posted:2000-series Threadripper doesn't have official ECC support, right? I wonder if the existence of X and WX variants of the same number indicates no-ECC/ECC support. Probably warranty/support/validation/out of band management. Bump the warranty and validate a variety of ECC sticks to work with the machine, so you have 'this is a factory HP/Dell/Lenovo box with ECC' factor going on. Toss in whatever the gently caress AMD is calling their version of vPro and you have a reason to pay an extra $300 for it.
|
|
#
?
Jul 31, 2018 00:32
|
|
|
|
| # ? Apr 28, 2024 03:58 |
|
|
Broose posted:Thread count so high you could use it as a thrifty blanket. Threadripper natively is worse than mainstream Ryzen in games, they have a "games mode" that basically halves the processor to keep it to one die and not demolish gaming performance. If you want gaming performance and don't mind dealing with the devil, wait and get a i9-9900K.
|
|
#
?
Jul 31, 2018 13:28
|
|
