|
Dr. Fishopolis posted:It's a clear, obvious, provable ~30% hit for iops with no roadmap for a better solution. The charts are rolling in in this very thread. This is a giant problem for datacenters. I’m pretty sure he doesn’t but he also doesn’t own any data centers.
|
# ? Jan 6, 2018 20:12 |
|
|
# ? Apr 29, 2024 09:37 |
|
https://www.phoronix.com/scan.php?page=article&item=linux-more-x86pti&num=1
|
# ? Jan 6, 2018 21:36 |
|
Dr. Fishopolis posted:It's a clear, obvious, provable ~30% hit for iops with no roadmap for a better solution. The charts are rolling in in this very thread. This is a giant problem for datacenters. Hell even if it were only 5% the large datacenters would care, from what I have read Google and the like will go to extreme lengths and great efforts to get small improvements in performance/watt even going as far as designing some of their own hardware.
|
# ? Jan 6, 2018 22:57 |
|
30% seems like an outlier. I don't have the link, but Red Hat's testing hit 10-12%, and I tend to trust that they know real-world scenarios for enterprise.
|
# ? Jan 6, 2018 23:38 |
|
Craptacular! posted:30% seems like an outlier. I don't have the link, but Red Hat's testing hit 10-12%, and I tend to trust that they know real-world scenarios for enterprise. Even that's terribad. That's like two to three CPU generations wiped out at a time when performance increases are hard to come by. Fame Douglas fucked around with this message at 01:10 on Jan 7, 2018 |
# ? Jan 7, 2018 01:04 |
|
Fame Douglas posted:Even that's terribad. That's like two to three CPU generations wiped out at a time when performance increases are hard to come by. Hey man, if you’ve got a 10% performance hit, well, you could throw more CPUs at it. I’d be happy to sell you more of what you already have, so you’ve got uniformity and whatnot...
|
# ? Jan 7, 2018 02:54 |
|
quote:On Xeon Scalable class hardware the I/O performance slowdowns when running the tests bare metal on the host tended to generally be just a few percent slower on affected benchmarks, less than with the "desktop" hardware. What? Any theories why that would be?
|
# ? Jan 7, 2018 04:02 |
|
Rastor posted:What? Any theories why that would be? Intel knew.
|
# ? Jan 7, 2018 04:32 |
|
I’m largely indifferent to all of this stuff going on, but I went ahead and updated my bios. The weird thing is, it was released by Asus on 1/4 which makes sense but when I go to load it in the BIOS it says 12/12/17. Hmmmm...
|
# ? Jan 7, 2018 07:37 |
|
If youre an end user, it really makes no difference. Having the latest BIOS is prolly a good thing though, especially if you have a x299 or z370 board, cause the early bios on those seemed pretty janky compared to the latest releases.
|
# ? Jan 7, 2018 08:27 |
|
VulgarandStupid posted:I’m largely indifferent to all of this stuff going on, but I went ahead and updated my bios. The weird thing is, it was released by Asus on 1/4 which makes sense but when I go to load it in the BIOS it says 12/12/17. Hmmmm... Companies knew. Intel found out in June, developers of Ubuntu Linux found out in November, I have to imagine ASUS is a little closer to the Intel side of the scale.
|
# ? Jan 7, 2018 08:32 |
|
I'm an end user, and the relevant windows update knocked 15 fps off World of Warships and World of Warplanes, having played it just before the update, and just after the update, with no other changes made. It's basically killed my i7 920 when I needed it to last another 4 months, so there's a greater than 100% chance I'll be getting a Ryzen 1700 or whatever the refresh is instead of anything Intel does.
|
# ? Jan 7, 2018 08:48 |
Enjoy playing your crappily optimised games on worst processors imaginable for them!
|
|
# ? Jan 7, 2018 08:50 |
|
At least they'll be playable, because the current 1-5fps isn't.
|
# ? Jan 7, 2018 08:53 |
NTRabbit posted:At least they'll be playable, because the current 1-5fps isn't. They are guaranteed to be playable at higher framerates on any comparable Intel CPU released in 2013 or later. But that's your money to waste so go ahead with your rage fit over end of life accident on a decade old CPU.
|
|
# ? Jan 7, 2018 08:56 |
|
cinci zoo sniper posted:They are guaranteed to be playable at higher framerates on any comparable Intel CPU released in 2013 or later. But that's your money to waste so go ahead with your rage fit over end of life accident on a decade old CPU. Interesting to categorise reaction to a major security flaw, criminal activity from the CEO, and a major loss of function due to the fix patch as a "rage fit", but whatever you need to feel good about your pom poms I guess
|
# ? Jan 7, 2018 09:00 |
NTRabbit posted:Interesting to categorise reaction to a major security flaw, criminal activity from the CEO, and a major loss of function due to the fix patch as a "rage fit", but whatever you need to feel good about your pom poms I guess Yea yea sod off to AMD thread or wherever trash unable to take criticism, like yourself, congregates. Probably even that thread will tell you that AMD sucks compared to Intel, with Meltdown degradation factored in, for single-core workloads which your entire "use" case. Big question is not the understandable loss of performance or Intel's actions, but why on earth are you using a 10 year old CPU to begin with. CPUs rarely last that much, on normal gaming workloads, and most components will give out even before that. cinci zoo sniper fucked around with this message at 09:06 on Jan 7, 2018 |
|
# ? Jan 7, 2018 09:04 |
|
No question Nehalem is getting long in the tooth for gaming but it's still really nice as a server, if you don't mind the power consumption. Right now my Z400 with a W3565 is my ZFS server and today I set up Emby to do transcoding, so I can watch movies across my lovely wifi without it stuttering like crazy
|
# ? Jan 7, 2018 09:04 |
|
cinci zoo sniper posted:Yea yea sod off to AMD thread or wherever trash unable to take criticism, like yourself, congregates. Probably even that thread will tell you that AMD sucks compared to Intel, with Meltdown degradation factored in, for single-core workloads which your entire "use" case. Nice meltdown
|
# ? Jan 7, 2018 09:05 |
PerrineClostermann posted:Nice meltdown I'm allergic to pom poms.
|
|
# ? Jan 7, 2018 09:07 |
|
If I had a use for a home server I might do that, but I really only have a use for a single desktop. My 920 has been soldiering on for a long time now, because I've not been able to afford an upgrade, and really still can't. The timing of this mega fuckup from Intel hasn't been great for me.cinci zoo sniper posted:Yea yea sod off to AMD thread or wherever trash unable to take criticism, like yourself, congregates. Probably even that thread will tell you that AMD sucks compared to Intel, with Meltdown degradation factored in, for single-core workloads which your entire "use" case. I'm sorry Mr President, I didn't realise a stable genius like yourself had an account
|
# ? Jan 7, 2018 09:08 |
NTRabbit posted:I'm sorry Mr President, I didn't realise a stable genius like yourself had an account PerrineClostermann posted:Nice meltdown Do let us know though how good ole BigWorld engine runs on a Ryzen.
|
|
# ? Jan 7, 2018 09:08 |
|
I will in a few months, when hopefully I'll be able to afford one. A 1700, an asrock board, and 16gb of DDR4-3200 is just under $1k dollarydoos, and it's not easy for me to put that much spare cash together. Still using the original cpu, motherboard, and memory from 2008, and I'd still be using the gfx card too but for the GTX 295 not supporting dx11 and me needing it to play Fallout 4. NTRabbit fucked around with this message at 09:15 on Jan 7, 2018 |
# ? Jan 7, 2018 09:10 |
|
Lets cut down on the shitposts / attacks guys, yeah? Re: BIOS build dates, usually that’s a compile-time token that gets populated when VeB runs the build. So yeah, that BIOS was built in December because Asus / others got briefed in earlier.
|
# ? Jan 7, 2018 09:13 |
|
Oh poo poo I didn't even think about it, my Coppermine Thinkpad is affected too Now it's personal!
|
# ? Jan 7, 2018 09:15 |
|
NTRabbit posted:I'm an end user, and the relevant windows update knocked 15 fps off World of Warships and World of Warplanes, having played it just before the update, and just after the update, with no other changes made. It's basically killed my i7 920 when I needed it to last another 4 months, so there's a greater than 100% chance I'll be getting a Ryzen 1700 or whatever the refresh is instead of anything Intel does. I doubt that has anything to do with the intel meltdown issue, honestly? That sounds like something else is wrong, like a driver not playing nice or somethin. e: NTRabbit posted:GTX 295 Oh poo poo, uh, yeah. I think you might be a collateral damage of an update where people were like 'the gently caress is GTX 295? why would i test that?' Thing is literally 9 years old. Cygni fucked around with this message at 09:25 on Jan 7, 2018 |
# ? Jan 7, 2018 09:21 |
Cygni posted:I doubt that has anything to do with the intel meltdown issue, honestly? That sounds like something else is wrong, like a driver not playing nice or somethin. It could easily be Meltdown-related, that CPU neither has PCID nor is fast as-is. Maybe its ancient enough for some extra PCID-style slowdowns.
|
|
# ? Jan 7, 2018 09:25 |
|
cinci zoo sniper posted:It could easily be Meltdown-related, that CPU neither has PCID nor is fast as-is. Maybe its ancient enough for some extra PCID-style slowdowns. I dunno man i aint seen anybody else report 93% frame rate decreases in games from these patches. I dont know poo poo but that seems like somethin else is up.
|
# ? Jan 7, 2018 09:26 |
|
Cygni posted:Oh poo poo, uh, yeah. I think you might be a collateral damage of an update where people were like 'the gently caress is GTX 295? why would i test that?' Thing is literally 9 years old. GTX 295 was what I first built this machine with, I'm currently running an R9 380, and I got the last driver update for that in December
|
# ? Jan 7, 2018 09:28 |
|
NTRabbit posted:GTX 295 was what I first built this machine with, I'm currently running an R9 380, and I got the last driver update for that in December ah gently caress im a dumb reader, sorry.
|
# ? Jan 7, 2018 09:29 |
Cygni posted:I dunno man i aint seen anybody else report 93% frame rate decreases in games from these patches. I dont know poo poo but that seems like somethin else is up. I'm not sure there are many people reporting BigWorld benchmarks for Bloomfield processors in January 2018, that'd be beyond bizarre.
|
|
# ? Jan 7, 2018 09:30 |
|
NTRabbit posted:I will in a few months, when hopefully I'll be able to afford one. A 1700, an asrock board, and 16gb of DDR4-3200 is just under $1k dollarydoos, and it's not easy for me to put that much spare cash together. If money is tight, I'd try to see if it's possible to uninstall the patch and see if that solves the performance issue, just to be 100% sure if it's just that patch that's the issue and not something else.
|
# ? Jan 7, 2018 10:04 |
|
Do you think they will have fixed it in architecture for the next Intel chips? Doesn't sound like they can get the performance back.
|
# ? Jan 7, 2018 13:02 |
|
Carecat posted:Do you think they will have fixed it in architecture for the next Intel chips? Doesn't sound like they can get the performance back. I read that existing designs in the pipeline will be affected too, apparently only a totally new CPU design with the intent to avoid the flaw will do the trick.
|
# ? Jan 7, 2018 14:33 |
|
Installed the 3601 (beta) update for the Asus H81M-P Plus. SpeculationControlSettings now report all green on the i5 4440. Don't know if the Asus update with microcode was necessary 'cause I didn't figure out how to run the SpeculationControl before now, but I needed to change the boot logo anyway
|
# ? Jan 7, 2018 15:23 |
|
I don't know if this is the right place to ask, but I just had that update for the fix yesterday, and today my RAM speed reverted to the default 2133 from 3000. I'm not too concerned but would the two possibly be related?
|
# ? Jan 7, 2018 16:46 |
|
Supreme Allah posted:I don't know if this is the right place to ask, but I just had that update for the fix yesterday, and today my RAM speed reverted to the default 2133 from 3000. I'm not too concerned but would the two possibly be related? Your bios dumped its settings when you updated the firmware. You just need to go in and select the xmp profile for your ram. I updated my bios to most recent from december, lost all my overclock settings, and it didn't fix spectre
|
# ? Jan 7, 2018 16:52 |
|
mewse posted:Your bios dumped its settings when you updated the firmware. You just need to go in and select the xmp profile for your ram. Ah thanks.
|
# ? Jan 7, 2018 16:55 |
|
PC Perspective got some interesting results from their post-Meltdown storage benchmarksPC Perspective posted:The 960 EVO comes in at that same 20% drop seen elsewhere, but check out the 850 EVO's nearly 10% *increase* in performance. The 900P pushes this further, showing an over 15% *increase*. You would figure that a patch that adds latency to API calls would have a noticeable impact on a storage device offering extremely low latencies, but that did not end up being the case in practice.
|
# ? Jan 7, 2018 17:32 |
|
|
# ? Apr 29, 2024 09:37 |
|
Getting BSODs since yesterday after installing the update. Sucks for me, but can’t imagine the headache we’ve still got ahead of us. And just as people are setting up the hardware they got for Christmas.
|
# ? Jan 7, 2018 17:38 |