Search Amazon.com:
Register a SA Forums Account here!
JOINING THE SA FORUMS WILL REMOVE THIS BIG AD, THE ANNOYING UNDERLINED ADS, AND STUPID INTERSTITIAL ADS!!!

You can: log in, read the tech support FAQ, or request your lost password. This dumb message (and those ads) will appear on every screen until you register! Get rid of this crap by registering your own SA Forums Account and joining roughly 150,000 Goons, for the one-time price of $9.95! We charge money because it costs us $3,400 per month for bandwidth bills alone, and since we don't believe in shoving popup ads to our registered users, we try to make the money back through forum registrations.
«105 »
  • Post
  • Reply
craig588
Nov 19, 2005



The 1060 only draws like 70 watts stock, which would be a lot for just the PCI-E slot, especially overclocked. The 1050 is about 50% smaller so uses even less power. Nvidia TDPs are the worst case scenario limits, limited by the bios which you usually only run into while overclocking. The TDP for the 1060 is 120 watts and the TDP for the 1050 is 75 watts, assuming the trend is kind of similar a 1050 probably uses about 40 watts stock, well within the PCI-E power limit.

Binning is all kind of a lie, especially for Pascal. Many Pascals are hitting nearly 2GHz just from the factory auto overclocking and the pre overclocked cards are only like 1,800MHz or so at the most. It's more like they'll give you a return if somehow it doesn't hit those speeds, but there's no real difference between the stock and factory overclocked version. (Most of the time, sometimes there are different PCBs or bigger fans on some models)

craig588 fucked around with this message at Aug 15, 2017 around 17:08

Adbot
ADBOT LOVES YOU

Zero VGS
Aug 16, 2002
"It has gunfights and shit!"


Lipstick Apathy

22 Eargesplitten posted:

How much power can a PCIe slot push by itself? I thought it was 75 watts, which would preclude the 1050ti from overclocking, but manufacturers make overclocked 1050tis.

The PCIe slot is only supposed to push 75 watts. The 1050ti is like the 750ti, where some are PCIe only, and some have supplemental power pins added, like this one:

https://www.pny.com/geforce-gtx-105...TX1050T4XGPB-OC

Edit:

It seems like some 1050ti's can still overclock a lot when bound to the 75 watts though:

http://www.legitreviews.com/overclo...-the-max_187385

22 Eargesplitten
Oct 10, 2010
I'm conducting an experiment to see if removing your shitty avatar will improve your shitty taste in black metal

Okay, thanks. I might have a line on a used MSI Armor 960, I'm considering that or a 1050ti. I figure the 960's warranty being out of warranty doesn't matter if I'm overclocking anyway, and it seems to have PCIe power connectors unless I'm reading specs for the wrong card. I just need to make sure it's the 4GB version.

craig588
Nov 19, 2005



If they're around the same price I'd get a 1050 ti every time. It generally performs better and uses less power. Power use doesn't matter a whole lot maybe a couple dollars a year, but if it's the same price no reason not to go with the lighter one. Also being on a more recent chip means you're more likely to get longer driver support.

22 Eargesplitten
Oct 10, 2010
I'm conducting an experiment to see if removing your shitty avatar will improve your shitty taste in black metal

Okay. I'll see if they are the same price, it seems like the 1050s just keep getting more expensive.

Would minis have a problem with cooling when overclocked?

Avalanche
Feb 2, 2007


Been playing around with my 6700k as I just figured out I've been overvolting it to 1.42 for the past year nice and stable (forgot to change in bios when I was screwing around with OCing when I did the initial build). Might as well run this little bastard into the ground.

What's weird is the chip is rock solid stable at 4.7ghz at 1.38 volts but is a little less stable at 4.7ghz at 1.42 volts.This is on a Noctua DH15S cooler.

Huh?

I thought these chips were supposed to crash hard if you didn't give them enough power?

Am I just running into thermal throttling issues? For example, I get get through intel burn test just fine at 1.38 with no cores shooting up to 99C or just plain turning off, but weird poo poo starts happening at 1.42 where temps will spike to 99C, processor cores will gray out/shut off on HWinfo, and the mobo will not recognize any usb or headphone inputs for a brief moment or just blue screen. Same thing with Cinebench where scores are actually a few points better when running at a lower voltage.

I've also noticed weird stuff going on in HWInfo when running at 1.42 volts where the Maximum Voltage will bump up WAY higher than the input maximum in the Asus board bios including all the way up to 1.47 or 1.48 volts. Same thing at 1.38 volts where there will be maximum reading spikes at 1.42, 1.44 volts, etc. Why the hell is this happening if the chip is set to run at a constant voltage? Is there some kind of voltage offset setting I'm forgetting to disable?

Should I try to push the voltage down further? Little worried about what could happen if I under volt the chip.

Zero VGS
Aug 16, 2002
"It has gunfights and shit!"


Lipstick Apathy

Avalanche posted:

Been playing around with my 6700k as I just figured out I've been overvolting it to 1.42 for the past year nice and stable (forgot to change in bios when I was screwing around with OCing when I did the initial build). Might as well run this little bastard into the ground.

What's weird is the chip is rock solid stable at 4.7ghz at 1.38 volts but is a little less stable at 4.7ghz at 1.42 volts.This is on a Noctua DH15S cooler.

Huh?

I thought these chips were supposed to crash hard if you didn't give them enough power?

Am I just running into thermal throttling issues? For example, I get get through intel burn test just fine at 1.38 with no cores shooting up to 99C or just plain turning off, but weird poo poo starts happening at 1.42 where temps will spike to 99C, processor cores will gray out/shut off on HWinfo, and the mobo will not recognize any usb or headphone inputs for a brief moment or just blue screen. Same thing with Cinebench where scores are actually a few points better when running at a lower voltage.

I've also noticed weird stuff going on in HWInfo when running at 1.42 volts where the Maximum Voltage will bump up WAY higher than the input maximum in the Asus board bios including all the way up to 1.47 or 1.48 volts. Same thing at 1.38 volts where there will be maximum reading spikes at 1.42, 1.44 volts, etc. Why the hell is this happening if the chip is set to run at a constant voltage? Is there some kind of voltage offset setting I'm forgetting to disable?

Should I try to push the voltage down further? Little worried about what could happen if I under volt the chip.

Temperature affects stability as well. At the higher voltage it might be heating itself up a few C extra, increasing it's own resistance and hence lowering stability.

GRINDCORE MEGGIDO
Aug 22, 2004

blub

1.5v seems pointless if you aren't getting 5ghz+ out of it. I'd set it to adaptive voltage, 4.8ghz, 1.4v boost and see if it plays nice, torch the current settings.

Khablam
Mar 29, 2012

I have always thought the actions of men the best interpreters of their thoughts.



You can typically get 4.7 from 1.35v - your issue is almost certainly overvoltage, not under.
There's almost entirely no risk from undervolting.

cat doter
Jul 27, 2006



gonna need more cheese...australia has a lot of crackers

So for the past few days I was running into stability issues with my old i5 3570k overclock, it was a pretty modest OC, 4.4ghz at 1.275v, but I'd randomly get hard freezes (and sometimes complete shutdowns) whenever my system was at idle. That's a really loving hard problem to solve, right? Why would it only ever crash at idle? I could play games that stressed the CPU for hours but never have an issue.

Thing is though the voltage mode was set to fixed, so that voltage was set pretty much constantly, and I've had some CPU stability issues in the past that was mostly fixed by increasing the voltage slightly. On a whim I decided to switch the voltage mode to offset. Now it's stopped crashing entirely.

Go loving figure. Everything I've read about overclocking has said that fixed is fine as long as you don't mind a bit more power draw.

Constellation I
Apr 3, 2005
I'm a sucker, a little fucker.

Maybe due to vdroop, your voltage at high load is actually lower than idle. And for some reason, your system hates the higher voltage even at no load? I dunno, hopefully you have no more issues though.

GRINDCORE MEGGIDO
Aug 22, 2004

blub

cat doter posted:

So for the past few days I was running into stability issues with my old i5 3570k overclock, it was a pretty modest OC, 4.4ghz at 1.275v, but I'd randomly get hard freezes (and sometimes complete shutdowns) whenever my system was at idle. That's a really loving hard problem to solve, right? Why would it only ever crash at idle? I could play games that stressed the CPU for hours but never have an issue.

Thing is though the voltage mode was set to fixed, so that voltage was set pretty much constantly, and I've had some CPU stability issues in the past that was mostly fixed by increasing the voltage slightly. On a whim I decided to switch the voltage mode to offset. Now it's stopped crashing entirely.

Go loving figure. Everything I've read about overclocking has said that fixed is fine as long as you don't mind a bit more power draw.

That's a weird one. I don't know but I got really weird voltage behavior on my Asus board with the rev 1 bios (offset voltage values did absolutely nothing). Maybe it's a bios issue?

cat doter
Jul 27, 2006



gonna need more cheese...australia has a lot of crackers

Constellation I posted:

Maybe due to vdroop, your voltage at high load is actually lower than idle. And for some reason, your system hates the higher voltage even at no load? I dunno, hopefully you have no more issues though.

It's completely stable now, hasn't crashed since. CPU-Z was saying that the voltage was slightly lower at idle, but I don't know if its reporting is 100% accurate.


GRINDCORE MEGGIDO posted:

That's a weird one. I don't know but I got really weird voltage behavior on my Asus board with the rev 1 bios (offset voltage values did absolutely nothing). Maybe it's a bios issue?

I really have no idea. The board appears to be up to date(it's an Asrock Z77 extreme 4), so if it's a bios issue I'm probably out of luck. Though I'm stable now so whatever!

cat doter
Jul 27, 2006



gonna need more cheese...australia has a lot of crackers

As a close to this story I was able to bump the frequency from 4.4ghz to 4.7ghz since the instability appears to indeed have been the motherboard bug. That's pretty high for an ivy bridge CPU.

22 Eargesplitten
Oct 10, 2010
I'm conducting an experiment to see if removing your shitty avatar will improve your shitty taste in black metal

A recommendation in the hardware thread was for me to overvolt since setting my Nvidia control panel settings to maximum performance actually made it worse. Is there a good guide to overvolting? I've never done anything other than slightly upping frequencies in Afterburner.

I've got a Gigabyte GTX970 OC Mini if it makes a difference.

craig588
Nov 19, 2005



The common wisdom is to not overvolt at all because it might raise you a bin, but it adds so much heat and power use for most cards it's not worth it and for a lot of people actually performs worse unless they have really good cooling to handle the extra heat. You sure they didn't mean raise the power, not the volts? As long as you've got the case airflow and power supply to deliver it maxing out the power target for every Nvidia card since Kepler is a very beneficial thing, almost improving performance directly proportionally to how much extra power you can give it.

You should be able to return the GPU offset clock to stock and max out the power target in Afterburner and not even have to stability test it.

22 Eargesplitten
Oct 10, 2010
I'm conducting an experiment to see if removing your shitty avatar will improve your shitty taste in black metal

Yep, my mistake. He mentioned more aggressive voltage in the context of the power setting, but suggested raising the power limit.

I don't have great airflow, but I'll keep an eye on temps using Rivatuner. Is there a non-torture test program that I should run to test how hot it will get?

Khablam
Mar 29, 2012

I have always thought the actions of men the best interpreters of their thoughts.



22 Eargesplitten posted:

Yep, my mistake. He mentioned more aggressive voltage in the context of the power setting, but suggested raising the power limit.

I don't have great airflow, but I'll keep an eye on temps using Rivatuner. Is there a non-torture test program that I should run to test how hot it will get?

Run a torture test.
Furmark will plot a temperature curve for you so you can see where it plateaus.

22 Eargesplitten
Oct 10, 2010
I'm conducting an experiment to see if removing your shitty avatar will improve your shitty taste in black metal

Will that stop when and if the card gets to an unsafe temperature?

craig588
Nov 19, 2005



Unless you flash the bios the card itself will stop itself from overheating. Furmark isn't a stability test, but it will test the worst case for power and cooling. It tests way too limited of a subset of the GPU to work as a stability test but it will let you know a worst case for power and temperature. You don't need to run it for a long time, it'll probably level out in less than 10 minutes, it makes heat like no game does.

22 Eargesplitten
Oct 10, 2010
I'm conducting an experiment to see if removing your shitty avatar will improve your shitty taste in black metal

I got a 12% power limit boost out of Afterburner got to 78 and only hit 76% fanspeed. Am I right in thinking that if I set up a more aggressive fan profile I could probably get a good bit more performance? Or am I likely to hit power issues if all I can get is 12%?

I'm not sure whether this is even relevant since I haven't tested in game performance yet, but I've been wondering this sort of stuff since I got the card.

craig588
Nov 19, 2005



Comparing power limits between cards is meaningless because they're arbitrary and set by the manufacturer. As Maxwell bioses are cracked you could make one that allows less power at 200% as another allows at 100%. (Pascals still aren't and at this point in the cycle are unlikely to be and it's a major hold back for overclocking) I always tell people to max it out though because it's a predefined limit from the manufacturer, nothing should go wrong unless the card already was faulty and should be warrantied.

Maxwells don't auto overclock nearly as well as Pascals, but speeding up the fan and cooling it down should still get you a bit more performance.

Adbot
ADBOT LOVES YOU

Zero VGS
Aug 16, 2002
"It has gunfights and shit!"


Lipstick Apathy

Every time I've adjusted my overclock, for years now, I've been wondering why my frequency has always been stuck at max and never speedsteps down. I got some advice on what to check but it never solved the issue.

Turns out I had my Energy Settings in Win 10 set to "High Performance" this whole time, and apparently that pegs the frequency. I thought it only affected HDD/Monitor/Sleep times.

On the other hand, now that my PC can throttle down to 800mhz from 4200mhz, it only seems to save like 3 watts, I'd have thought it would be a bigger deal.

  • 1
  • 2
  • 3
  • 4
  • 5
  • Post
  • Reply
«105 »