|
Methylethylaldehyde posted:It's the reason I got the Taichi instead. Plus I'm saving my lunch money for some QDR infiniband off ebay if I find a good deal. This is a good answer.
|
#
?
Feb 6, 2018 08:09
|
|
|
|
| # ? May 16, 2024 18:28 |
|
|
I’m on a Z170 but my Asrock’s E2400 Killer NIC can push 106MB/sec to my server. Given that it’s connected using salvaged-from-construction-site cat 5E patch cable I self-crimped and a crappy Adata 5 Port Gigabit Switch. I’m going to say it’s ok enough as long as you don’t touch the poxy software. In fact if I recall correctly, I went and manually extracted the INF, etc. out of the driver zip file and installed via add new hardware > have disk... because I didn’t want any of the software near my PC.
|
|
#
?
Feb 6, 2018 15:39
|
|
|
Nam Taf posted:In fact if I recall correctly, I went and manually extracted the INF, etc. out of the driver zip file and installed via add new hardware > have disk... because I didn’t want any of the software near my PC. There it is! I did this and now it works well enough, the intel nic is still slightly better in my not at all scientific testing. Thank you I haven't seen that Add New Hardware dialog in a long rear end time. It's loving weird that I have nostalgia for various windows dialogs
|
|
#
?
Feb 6, 2018 16:22
|
|
|
I don't understand the "MSI doesn't update" thing. I have an MSI B350 and I've been getting updates consistently over the last 12 months. Do other companies pump out updates? I never remember updating my old ASRock Z77 more than twice, but that was Intel hardware so maybe that's not fair to compare it to a newer and less established Ryzen platform.
|
|
#
?
Feb 6, 2018 17:50
|
|
|
Tiny Tubesteak Tom posted:I don't understand the "MSI doesn't update" thing. I have an MSI B350 and I've been getting updates consistently over the last 12 months. Do other companies pump out updates? I never remember updating my old ASRock Z77 more than twice, but that was Intel hardware so maybe that's not fair to compare it to a newer and less established Ryzen platform. Yeah, other companies do update more frequently. Asus has pushed probably 5 BIOS updates since September/October, which is the last time 90% of MSI's AM4 boards got updated. Edit: Holy poo poo, MSI heard us talking about them. They released BIOS updates for most of their AM4 boards today. People on the AMD reddit board are reporting much better memory overclocking capability compared to last BIOSes, the ability to disable the PSP, and also they fixed the CoolNQuiet bug I mentioned up a few posts. MSI delivers! bobfather fucked around with this message at 19:56 on Feb 6, 2018 |
|
#
?
Feb 6, 2018 18:14
|
|
|
https://www.extremetech.com/computing/263286-sitting-globalfoundries-talk-7nm-euv Info about Glofo 7nm and the EUV transition
|
|
#
?
Feb 6, 2018 22:16
|
|
|
[H] opened up the floodgates. Legit Reviews and HotHardware have Threadripper mining articles now. How much longer before PC World/Gamer picks up on it?  Please hurry up and crash faster, buttcoin. SwissArmyDruid fucked around with this message at 03:07 on Feb 7, 2018 |
|
#
?
Feb 7, 2018 03:03
|
|
|
New Zealand can eat me posted:I wouldn't worry too much between brands, IMO the most important thing is the number of VRM phases. There's a decent temperature/stability difference between 8 and 10/12 phase This isn't even something a lot of manufacturers disclose, but from what I can tell all the motherboards I'm looking at (A320, we're never OCing this HTPC and besides I've never once OCed any PC even my unlocked gaming rigs) all have 6 or so, and 9 in the case of the ASRock Pro. What kind of "stability" are we talking about, because this is a machine that's running on stock clocks 99% of the year? karoshi posted:They were (are?) a favorite for hackintosh builds. This hasn't been the case since Clover matured enough that we could finally say no one should be using Chameleon anymore. Craptacular! fucked around with this message at 09:52 on Feb 7, 2018 |
|
#
?
Feb 7, 2018 09:49
|
|
|
I'd look at one with 12 phases if I was thinking of dropping a new chip into the board at some point, because they might release one eventually with higher power requirements. I kind of doubt they'd release one that needed more phases to be stable, but who knows when AMD.
|
|
#
?
Feb 7, 2018 14:10
|
|
|
Craptacular! posted:This isn't even something a lot of manufacturers disclose, but from what I can tell all the motherboards I'm looking at (A320, we're never OCing this HTPC and besides I've never once OCed any PC even my unlocked gaming rigs) all have 6 or so, and 9 in the case of the ASRock Pro. The A320 Pro is a 3+3 Phase VRM with doubled up VCores (You can usually just count the beefy rectangles to the left of the CPU socket afaik). I wouldn't worry about it if you aren't going to overclock or undervolt
|
|
#
?
Feb 7, 2018 16:11
|
|
|
Yeah, you guys aren't exactly talking about stuff that is easily explained in motherboard marketing. I've built a bunch of computers in my time (admittedly only one AMD back in the stone age) and never thought about VRMs. But I've never run a single CPU outside of stock and have thrown away "free performance" from every CPU I've owned over the past 20 years because it's 'not worth the risk'. My current aging setup I gave Intel the extortion money they wanted for a K chip and have run it at stock speeds for five years because, again, why risk. I just want to put the upcoming APU in a motherboard that will run always-on. I ask about stability because I haven't owned AMD since the K6-2 and that machine was unstable precisely because of it's motherboard. My understanding is that the true hell I'm going to experience is that nothing I buy in a store will reach POST with the chip in it because it's Too drat New and requires a BIOS update to, well, even update the BIOS. Right now I'm waiting for some motherboard manufacturer to confidently say "yes, Raven Ridge will boot under our boards."
|
|
#
?
Feb 7, 2018 19:27
|
|
|
....Modern overclocking is a risk? I thought it was just clever marketing for higher clocked chips. "Oh, you're totally pushing it to the edge! You're the one bringing the chip to its maximum potential! Isn't that amazing and worth 100 dollars?"
|
|
#
?
Feb 7, 2018 19:29
|
|
|
Hot take: VRM's really don't matter much unless you are running big voltage increases, or the chip needs a ton of power just at stock (like an x299 or x399) For a normal overclock with a normal desktop CPU, I haven't found it to make a huge difference at all. You end up hitting other limiting factors before the VRM most of the time. I may be wrong, but thats my personal experience.
|
|
#
?
Feb 7, 2018 19:42
|
|
|
PerrineClostermann posted:....Modern overclocking is a risk? I thought it was just clever marketing for higher clocked chips. "Oh, you're totally pushing it to the edge! You're the one bringing the chip to its maximum potential! Isn't that amazing and worth 100 dollars?" The real trick is that it allows chip manufacturers to benefit from performance numbers that they don't actually have to support. Pretty much all coffee lake CPUs should make it to an all-core 5 GHz... But if they don't, it's not Intel's problem
|
|
#
?
Feb 7, 2018 20:00
|
|
|
PerrineClostermann posted:....Modern overclocking is a risk? I thought it was just clever marketing for higher clocked chips. "Oh, you're totally pushing it to the edge! You're the one bringing the chip to its maximum potential! Isn't that amazing and worth 100 dollars?" I'm old, I remember when HardOCP was a young site and OC was definitely not-encouraged, not-marketed, and a way for some broke-rear end college students to circumvent a bottleneck and "stick it to the man". It was once this "Cool Kid found unbelievable way to get free performance! Intel hates him!" thing that you did because you were poor, tech was expensive, and you valued performance-for-money over warranty or stability or the thing still booting in 18 months. That's how I still see it. Much like smoking weed and using Linux, it's gone mainstream and corporate, but I still avoid it.
|
|
#
?
Feb 7, 2018 20:11
|
|
|
Craptacular! posted:I'm old, I remember when HardOCP was a young site and OC was definitely not-encouraged, not-marketed, and a way for some broke-rear end college students to circumvent a bottleneck and "stick it to the man". It was once this "Cool Kid found unbelievable way to get free performance! Intel hates him!" thing that you did because you were poor, tech was expensive, and you valued performance-for-money over warranty or stability or the thing still booting in 18 months. I mean if you're passing up literally free performance (since you still buy the unlocked chips...) that's on you, lol
|
|
#
?
Feb 7, 2018 20:14
|
|
|
SSJ_naruto_2003 posted:I mean if you're passing up literally free performance (since you still buy the unlocked chips...) that's on you, lol Yeah. This is a really odd attitude to take.
|
|
#
?
Feb 7, 2018 20:25
|
|
|
Cygni posted:Hot take: VRM's really don't matter much unless you are running big voltage increases, or the chip needs a ton of power just at stock (like an x299 or x399) Yeah, same. I've been happy with MSI for quite some time because even though they use a lot of cheap 50-cent power components, it only matters when you start drawing way more power than the stock CPU they're targeting. I'll take the easy 10% OC or whatever is available with stock voltage, which is what the average MSI board is good for. As soon as you start pumping extra volts and 150 watts through them they're crap, or at least that's what der8aurer says. I wouldn't know, I don't put 150 watts into my CPU. SSJ_naruto_2003 posted:I mean if you're passing up literally free performance If I really needed the extra performance I'd just spend the money for it (OTOH I also stopped buying unlocked intel chips because the value wasn't there anymore. I have a 1600X now, but when I bought it the 1600 and X were the same price.)
|
|
#
?
Feb 7, 2018 21:11
|
|
|
By that you mean change a 43 to a 50 and forget about it?
|
|
#
?
Feb 7, 2018 21:15
|
|
|
Klyith posted:Yeah, same. I've been happy with MSI for quite some time because even though they use a lot of cheap 50-cent power components, it only matters when you start drawing way more power than the stock CPU they're targeting. I'll take the easy 10% OC or whatever is available with stock voltage, which is what the average MSI board is good for. Uhh most of the time you literally click an auto overclock button ime. You can push it further but you don't have to.
|
|
#
?
Feb 7, 2018 21:21
|
|
|
Craptacular! posted:I'm old, I remember when HardOCP was a young site and OC was definitely not-encouraged, not-marketed, and a way for some broke-rear end college students to circumvent a bottleneck and "stick it to the man". It was once this "Cool Kid found unbelievable way to get free performance! Intel hates him!" thing that you did because you were poor, tech was expensive, and you valued performance-for-money over warranty or stability or the thing still booting in 18 months. Hardocp was the poo poo back then, with a lot of real experimentation going on. Loved my dual celeron BP6 setup, it really had some staying power. taqueso fucked around with this message at 23:29 on Feb 7, 2018 |
|
#
?
Feb 7, 2018 22:02
|
|
|
taqueso posted:Hardocp was the poo poo back then, with a lot of real experimentation going on. Loved by dual celeron BP6 setup, it really had some staying power. I'll never forget their expose and investigation into the Infinium Phantom.  In retrospect, their concept was ahead of its time.
|
|
#
?
Feb 7, 2018 22:32
|
|
|
PerrineClostermann posted:Yeah. This is a really odd attitude to take. Meh, I just want my machine to be stable. If the overclock is stable for you, that is great, but I don't want to have unexplained errors while debugging some software or my machine crapping out when rendering stuff. Even before I'd rather pay some more money for the peace of mind. YMMV ofcourse.
|
|
#
?
Feb 7, 2018 22:34
|
|
|
Mr Shiny Pants posted:Meh, I just want my machine to be stable. I hope you're not running Windows or an nVidia GPU, then.
|
|
#
?
Feb 7, 2018 22:39
|
|
|
PerrineClostermann posted:I hope you're not running Windows or an nVidia GPU, then. why...?
|
|
#
?
Feb 7, 2018 22:54
|
|
|
Mr Shiny Pants posted:Meh, I just want my machine to be stable. If the overclock is stable for you, that is great, but I don't want to have unexplained errors while debugging some software or my machine crapping out when rendering stuff. While I can understand the belief that an overclock might be unstable, most overclocks are tested to be stable at the speed they're running at. That's the point of doing tests at particular speeds and voltages, to determine if your OC is stable. Once stable it tends to remain that way for years. While I've heard of some older chips becoming less stable at high OCs over time, I haven't experienced that myself in anything from my celeron 300a that ran at 450mhz to my i5-4670K. Maybe it will happen some day but not so far. For maximum stability you'd probably want a server chip that supports ECC memory. Memory errors are far more common than you'd think. https://www.youtube.com/watch?v=lZ8s1JwtNas
|
|
#
?
Feb 7, 2018 22:57
|
|
|
PerrineClostermann posted:I hope you're not running Windows or an nVidia GPU, then. Windows is pretty solid when rendering and Visual Studio is also pretty stable. At least on my machines, anecdotal and all I know but it is what it is. Rexxed posted:While I can understand the belief that an overclock might be unstable, most overclocks are tested to be stable at the speed they're running at. That's the point of doing tests at particular speeds and voltages, to determine if your OC is stable. Once stable it tends to remain that way for years. While I've heard of some older chips becoming less stable at high OCs over time, I haven't experienced that myself in anything from my celeron 300a that ran at 450mhz to my i5-4670K. Maybe it will happen some day but not so far. I had my 566 Celeron clocked to 850 and some other chips too.  What would a TR1950X overclock to? I am running it stock now on a X399 Taichi with a Kraken X62. Just curious. I have 3200 MHZ memory from G-Skill.
|
|
#
?
Feb 7, 2018 23:02
|
|
|
The only time I thought I was experiencing electromigration, it ended up being caps on their way out. The voltage I bumped ended up being enough to make one finally barf and let me know what was up Cygni posted:Hot take: VRM's really don't matter much unless you are running big voltage increases, or the chip needs a ton of power just at stock (like an x299 or x399) It's not about max power, the additional phases allow you to run higher clocks at lower voltages. You're generally not going to see a B350 board get to 4100mhz without entering the Danger Zone of >1.5v w/ 1.25v SOC. (1.4 w/ peaks of 1.45 and 1.2 soc max are the official safe 24/7 numbers from AMD) I don't think there's anyone left trying to run these boards like this. This X370 Gaming 5 is doing the same clockspeed down at 1.375v w/ 1.15v soc
|
|
#
?
Feb 7, 2018 23:10
|
|
|
Mr Shiny Pants posted:Windows is pretty solid when rendering and Visual Studio is also pretty stable. At least on my machines, anecdotal and all I know but it is what it is. Some articles suggest 4 or 4.1ghz with very extreme cooling. I suspect 3.8 or 3.9 would be more reasonable with an AIO cooler but I'm sure there will be variance from chip to chip as to how much voltage you'd need and how much power it would draw.
|
|
#
?
Feb 7, 2018 23:11
|
|
|
You should be able to get to 4100 with ~1.2875v or so. Start at 1.2 and work your way up until it's stable. The best I've seen anyone manage is 1.225v, essentially winning the silicon lotto To get to 4200, you'd need to go all the way past 1.4v. Don't do this if you're not prepared to deal with >330w of heat
|
|
#
?
Feb 7, 2018 23:26
|
|
|
Kids these days with their one-click oc’s? Bah! REAL overclocking involved using a pencil on your Duron.
|
|
#
?
Feb 8, 2018 00:50
|
|
|
I'll see what it does when I have some time. Thanks guys.
|
|
#
?
Feb 8, 2018 09:06
|
|
|
Craptacular! posted:Yeah, you guys aren't exactly talking about stuff that is easily explained in motherboard marketing. I've built a bunch of computers in my time (admittedly only one AMD back in the stone age) and never thought about VRMs. But I've never run a single CPU outside of stock and have thrown away "free performance" from every CPU I've owned over the past 20 years because it's 'not worth the risk'. My current aging setup I gave Intel the extortion money they wanted for a K chip and have run it at stock speeds for five years because, again, why risk. Speak of the devil: https://videocardz.com/74956/asrock-announces-bios-support-for-amd-desktop-ryzen-2000-series
|
|
#
?
Feb 8, 2018 11:00
|
|
|
SwissArmyDruid posted:Speak of the devil: They were pumping updated out frequently for that mobo until Sept of last year and then they just stopped. Maybe had something to do with trying to come up with a Spectre fix?
|
|
#
?
Feb 8, 2018 20:21
|
|
|
SSJ_naruto_2003 posted:I mean if you're passing up literally free performance (since you still buy the unlocked chips...) that's on you, lol PerrineClostermann posted:Yeah. This is a really odd attitude to take. The 3770K I'm still using was like $20 more than the 3770 when I bought it in December 2012. Even though I have never used OC (and now it's attached to an H80 that's been running on Low-Noise for five years and I'm afraid cranking it up at this point might break something), it's nice that it's still there. If the chip had to have been sold by now, it would have helped it's resale value. I looked into OC last month. The cooler went from low to medium without much issue, but the Meltdown patch appears to have broken ASUS Digi Suite II and as their current mobos are on Suite III they have little interest in going back and fixing it. Since I don't like manually adjusting things in BIOS and would rather have software apply a "safe" OC I'm just sort of back to indifference again. EDIT: Apparently Intel makes a tool called XTU that does something similar. Maybe theirs is updated still. Thanks thread for giving me something to research on a lazy afternoon. SwissArmyDruid posted:Speak of the devil: This is kind of what I was afraid of. It sounds like the boards produced before very literally right now won't even boot to allow you to do the BIOS update. You have to own a Ryzen 2017 just to get the update done, and I don't have one. Craptacular! fucked around with this message at 22:14 on Feb 8, 2018 |
|
#
?
Feb 8, 2018 22:03
|
|
|
XTU is still maintained. I use it to dial in stable clocks and then just manually set those values in bios so they're persistent and not dependent on some janky software suite
|
|
#
?
Feb 8, 2018 22:25
|
|
|
Craptacular! posted:This is kind of what I was afraid of. It sounds like the boards produced before very literally right now won't even boot to allow you to do the BIOS update. You have to own a Ryzen 2017 just to get the update done, and I don't have one. Yeah, sadly, it seems that BIOS Flashback is only a feature available on ASRock's X399 boards. (or it seems, almost exclusively any motherboard manufacturer's top-end boards) Which is a little dumb, because that would solve any of your BIOS concerns if they brought it down to the X370 boards at least.
|
|
#
?
Feb 8, 2018 22:58
|
|
|
BangersInMyKnickers posted:XTU is still maintained. I use it to dial in stable clocks and then just manually set those values in bios so they're persistent and not dependent on some janky software suite XTU actually does its OCs as a lightweight system service so they get applied when Windows boots and doesn't need the big ugly software to be open at all. Pretty sure they're applied well before logon.
|
|
#
?
Feb 8, 2018 23:06
|
|
|
It might be because my board is old, but the vcore offset adjustments I was making weren't sticking with XTU
|
|
#
?
Feb 9, 2018 00:12
|
|
|
|
| # ? May 16, 2024 18:28 |
|
|
PC LOAD LETTER posted:Nice to see the X370 Taichi finally get a new AGESA version update too by Fri. As I understand it, each early release added features but added a bug. The current 3.20 has some core clock locking bug where the processor would be stuck at 1.55Ghz despite whatever settings you applied. There’s some beta BIOS floating around that fixes it but I’m sure as hell not comfortable downloading some BIOS update file from some random place. I just left my overclock at a modest setting until it would come out. Looking forward to tomorrow.
|
|
#
?
Feb 9, 2018 01:15
|
|
