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 money per month for bills, and since we don't believe in showing ads to our users, we try to make the money back through forum registrations.
 
  • Post
  • Reply
cstine
Apr 15, 2004

What's in the box?!?

Nomyth posted:

What is the gun he's open carrying

Is that a 10/22

lmao he can't even loving commit, and pack a real centerfire rifle round

Itís an airsoft rifle with the orange tip removed.

Adbot
ADBOT LOVES YOU

AlternateAccount
Apr 25, 2005
FYGM


Rude :(

K8.0
Feb 26, 2004

Her Majesty's 56th Regiment of Foot

GN continues to go scorched earth on Gigabyte

Klyith
Aug 3, 2007

GBS Pledge Week


Hey guys, guys,
*rips bong*
did you know... wait this is crazy...

Did you know, that if you press play on Dark Side of the Moon at the same time as Patrick presses start on the PSU test, that the music syncs up perfectly? Except you only get 3 songs into the album.

mearn
Aug 2, 2011

Kevin Harvick's #1 Fan!



I just got an email from Newegg re: Gigabyte PSUs. It looks like they're going to be issuing refunds or replacements directly, which is nice because I have no desire to deal with Gigabyte's RMA system again, nor do I want another Gigabyte PSU for the next decade or so.

Dr. Video Games 0031
Jul 16, 2004



GN Steve reigns victorious once again.

Wild EEPROM
Jul 29, 2011


oh, my, god. Becky, look at her bitrate.


[wr] psu test speedrun any%

gradenko_2000
Oct 5, 2010

Ask me about mapping out all the best limousine routes in Moscow for you and the little miss ;)

Lipstick Apathy

I kind of like how laidback the atmosphere in the GN studio seems to be, what with wearing shorts and slippers and all, but also that Steve isn't the only one with a good camera presence and confidence

cage-free egghead
Mar 8, 2004


Looks like they're Rainbow sandals, not slippers. :q:

I only know because I wear the same things lol

I do wonder what the long play is going to be for Gigabyte after this because this should make people scrutinize everything they put out, not just power supplies.

jisforjosh
Jun 6, 2006

"It's J is for...you know what? Fuck it, jizz it is"


cage-free egghead posted:

Looks like they're Rainbow sandals, not slippers. :q:

I only know because I wear the same things lol

I do wonder what the long play is going to be for Gigabyte after this because this should make people scrutinize everything they put out, not just power supplies.

Yeah, making me side eye my motherboard but I haven't had a single issue with a Gigabyte motherboard (my last 3). You'd have to pay me to touch an MSI or Asus board after the experiences I've had with them

Chevy Slyme
May 2, 2004

We're Gonna Run.

We're Gonna Crawl.

Kick Down Every Wall.


jisforjosh posted:

Yeah, making me side eye my motherboard

This but my GPU

mewse
May 2, 2006



I've been using gigabyte stuff since the core 2 duo era which was just after the capacitor plague and gigabyte advertised "ultra durable" motherboards with 100% japanese capacitors. My current Z370 motherboard is from gigabyte, and dealing with their bios during overclocking had me admitting to myself that they're just another OEM.

It's just unfortunate that they'd let their reputation drop like this to save money on a recall. GN is being pretty open about their testing methodology and surely gigabyte can reproduce these failures in a lab. They have an argument that the test is synthetic, not reflecting real world usage, but the PSU should fail gracefully rather than exploding and for a company that wants a reputation for reliability they are losing a lot of trust.

Pilfered Pallbearers
Aug 2, 2007




mewse posted:

I've been using gigabyte stuff since the core 2 duo era which was just after the capacitor plague and gigabyte advertised "ultra durable" motherboards with 100% japanese capacitors. My current Z370 motherboard is from gigabyte, and dealing with their bios during overclocking had me admitting to myself that they're just another OEM.

It's just unfortunate that they'd let their reputation drop like this to save money on a recall. GN is being pretty open about their testing methodology and surely gigabyte can reproduce these failures in a lab. They have an argument that the test is synthetic, not reflecting real world usage, but the PSU should fail gracefully rather than exploding and for a company that wants a reputation for reliability they are losing a lot of trust.

They donít have this excuse.

Iíll choose to believe GN that these are tested similarly at factory level.

Additionally, they tested a bunch of other PSUs the same way and they didnít fail, let alone explode. If they just failed from this testing, GN wouldnít be pushing so hard, but they loving explode under reproducible conditions. That should never happen.

TheFluff
Dec 13, 2006

FRIENDS, LISTEN TO ME
I AM A SEAGULL
OF WEALTH AND TASTE


The super high OPP threshold is also sort of concerning in and of itself even disregarding the component failures - even some absolute bargain basement 80+ bronze PSU's like the corsair cx series has quite reasonable OPP that consistently trips at like 120% load, at least on the 12V rails. Sometimes they skimp out hard on OPP on the minor rails but that's less of a big deal since there really isn't much in the way of consumption there.

GutBomb
Jun 15, 2005

Dude?


mearn posted:

I just got an email from Newegg re: Gigabyte PSUs. It looks like they're going to be issuing refunds or replacements directly, which is nice because I have no desire to deal with Gigabyte's RMA system again, nor do I want another Gigabyte PSU for the next decade or so.

I wonder if this will apply to their in-house prebuilt brand ABS. My dad just bought an ABS system from newegg a month ago that had one of these PSUs in it.

Klyith
Aug 3, 2007

GBS Pledge Week


mewse posted:

They have an argument that the test is synthetic, not reflecting real world usage,

Nah, that part is in fact the biggest bullshit of their statement, and one that only PC nerds would swallow because their main experience is PC benchmarking. But a PSU test isn't benchmarking a video card with a real game vs 3dmark.


Check this totally synthetic testing:
https://www.youtube.com/watch?v=Ai2HmvAXcU0

That's how real engineers test things. You test to destruction with controlled conditions, while carefully watching how / where / why the failure happens. If you find a replicable weakness that could produce failure in real conditions, you fix it. Gigabyte is making a plane where the wings fold in half as soon as you start that test, insisting that it's not real world failure. (And meanwhile, their plane seems to be crashing a whole lot when the wings mysteriously fall off mid-flight. Ignore that man behind the curtain!)


Gigabyte's bullshit about the testing being wrong is probably what Steve is the most mad about, and why he went full Hulk Smash on them with a dedicated episode for killing the PSU live on video. That's getting into the category of not just PR truth-bending, but telling dangerous lies. It's like a medical company being anti-vax about Covid or something.

CerealKilla420
Jan 2, 2014

"I need a handle man..."

gradenko_2000 posted:

That isn't quite the worst thing to know about the 8 Bit Guy

https://twitter.com/RealTimeKodi/status/1309886976837980161?s=19

Really? Is there some kind of 'me too' sort of stuff going on here?

I've never been able to watch the guy because he looks like a sludge monster and has the posture of a turtle, but he certainly looks like the kind of guy that would be a sex offender registry so nothing would really surprise me.

Raygereio
Nov 12, 2012


mewse posted:

They have an argument that the test is synthetic, not reflecting real world usage
Gigabyte doesn't have an argument there. Their press statement was just nonsense.

A power supply does not care that the current being drawn by a load tester, or by a GPU. If anything normal usage of the power supply in your system would be more stressful to it because the load would constantly vary.

K8.0
Feb 26, 2004

Her Majesty's 56th Regiment of Foot

Nah that's not how it works. The advanced features of the GP-P750GM can detect a synthetic load. Upon detecting said load, the PSU enters "blow the gently caress up" mode and detonates at a random time. This is good design because

Warmachine
Jan 30, 2012





Klyith posted:

Nah, that part is in fact the biggest bullshit of their statement, and one that only PC nerds would swallow because their main experience is PC benchmarking. But a PSU test isn't benchmarking a video card with a real game vs 3dmark.


Check this totally synthetic testing:
https://www.youtube.com/watch?v=Ai2HmvAXcU0

That's how real engineers test things. You test to destruction with controlled conditions, while carefully watching how / where / why the failure happens. If you find a replicable weakness that could produce failure in real conditions, you fix it. Gigabyte is making a plane where the wings fold in half as soon as you start that test, insisting that it's not real world failure. (And meanwhile, their plane seems to be crashing a whole lot when the wings mysteriously fall off mid-flight. Ignore that man behind the curtain!)


Gigabyte's bullshit about the testing being wrong is probably what Steve is the most mad about, and why he went full Hulk Smash on them with a dedicated episode for killing the PSU live on video. That's getting into the category of not just PR truth-bending, but telling dangerous lies. It's like a medical company being anti-vax about Covid or something.


Raygereio posted:

Gigabyte doesn't have an argument there. Their press statement was just nonsense.

A power supply does not care that the current being drawn by a load tester, or by a GPU. If anything normal usage of the power supply in your system would be more stressful to it because the load would constantly vary.

It's this. As Raygereio, Klyith, and Steve himself have said, a synthetic load is less stressful since you aren't dealing with transient currents and fluctuating loads like you are in a real system performing a workload, then waiting for a new load, then working, then waiting, then working some more.

GN is collecting scalps at this point, and I look forward to the next report because drat I wanna see Gigabyte try and weasel out before someone at the CPSC notices.

FuturePastNow
May 19, 2014




64bit_Dophins posted:

Really? Is there some kind of 'me too' sort of stuff going on here?

I've never been able to watch the guy because he looks like a sludge monster and has the posture of a turtle, but he certainly looks like the kind of guy that would be a sex offender registry so nothing would really surprise me.

As far as I know, he's only an offender to common sense

mewse
May 2, 2006



Holy poo poo that airplane wing EXPLODED, I'd never fly in that plane

mA
Jul 9, 2001
I am the ugly lover.

Good luck to anyone RMAing that gigabyte PSU (or any gigabyte compnent). Apparently the recent Gigabyte hack has really taken a toll on their RMA service, which was already the worst in the business beforehand.

Space Racist
Mar 27, 2008

~savior of yoomanity~


jisforjosh posted:

Yeah, making me side eye my motherboard but I haven't had a single issue with a Gigabyte motherboard (my last 3). You'd have to pay me to touch an MSI or Asus board after the experiences I've had with them

Iíve only used Gigabyteís motherboards thus far, and my experience has always been that of solid hardware marred by subpar BIOSes.

With these PSUs, suddenly Iím doubting their hardware as well. I hope their PR department recognizes the long term reputational damage theyíre incurring before they push back any further against GN et al.

Theophany
Jul 22, 2014

#blessed



Space Racist posted:

Iíve only used Gigabyteís motherboards thus far, and my experience has always been that of solid hardware marred by subpar BIOSes.

With these PSUs, suddenly Iím doubting their hardware as well. I hope their PR department recognizes the long term reputational damage theyíre incurring before they push back any further against GN et al.

Gigabyte have always made solid mid-tier boards imo. Before I started dropping shithead money on motherboards they were always my go to and I was never disappointed. Same with DFI but well, RIP.

After the 16MB ROM fiasco with MSI.... just lmao at ever buying their loving trash again at any price point.

Sidesaddle Cavalry
Mar 15, 2013


It's impossible to judge whether or not your previous hardware is bad based solely on current events. Just let the scandal guide your future purchases

Canine Blues Arooo
Jan 7, 2008

when you think about it...i'm the first girl you ever spent the night with



Grimey Drawer

KozmoNaut posted:

At least wear clothes that fit instead of some formless baggy mess and shorts with gigantically wide legs that go beneath the knees.

It's not that hard to buy a decent pair of jeans or well-cut shorts and a short sleeved shirt or polo shirt. Maybe you won't be the height of fashion, but at least you won't look like you just grabbed something at random from the closet, without turning on the lights.

My entire wardrobe is loose polyester tshirts and long, wide shorts. That poo poo is comfy and easy to wear, and I'm long past the point of caring what people think :P.

Klyith
Aug 3, 2007

GBS Pledge Week


Warmachine posted:

It's this. As Raygereio, Klyith, and Steve himself have said, a synthetic load is less stressful since you aren't dealing with transient currents and fluctuating loads like you are in a real system performing a workload, then waiting for a new load, then working, then waiting, then working some more.

So I wanna follow up with one clarification about this, because I think Steve's kinda tossed-off comment has the potential to be misunderstood. (It's the one thing I didn't like in that vid, but he's not working from script so it's hard to say things in ways that are both comprehensible to the audience and fully accurate.)

Mosfets are turning on and off all the time; that's how they work. The thing that produces 'wear' on them in a PSU is current and heat*. The main reason that a fluctuating load produces more wear in a normal scenario is that the transient peaks are much worse for them than the valleys are good. I am sure that a PSU held at 50% load will last longer than one that's bouncing between 10% and 90% while averaging 50%. However, I'm equally sure that holding it steady at 90% will kill it much faster than either.

*most of the time the current & heat are the same thing, because more current = more heat. but when they're different, in i dunno a LN2-cooled PSU, over current still kills.


So TBQH driving the mosfets at >130% of their rated load is close to the worst thing you can do to them, which is why that example was dead after 2 minutes of torture. If their testing procedure involved twiddling the buttons on their load tester to simulate a real-world load, it probably wouldn't kill them as fast.

But that would be stupid because the whole point of the test was to see if the OCP on those PSUs is set to a consistent and reasonable point (no) and whether it does the job of saving the PSU from catastrophic failure (no).



(I'm writing :words: just to make sure that nobody gets ideas like "mining bitcoin will make my PC last longer because steady state load is better than fluctuating!")

Klyith fucked around with this message at 22:07 on Aug 25, 2021

Canine Blues Arooo
Jan 7, 2008

when you think about it...i'm the first girl you ever spent the night with



Grimey Drawer

Klyith posted:

So TBQH driving the mosfets at >130% of their rated load is close to the worst thing you can do to them, which is why that example was dead after 2 minutes of torture. If their testing procedure involved twiddling the buttons on their load tester to simulate a real-world load, it probably wouldn't kill them as fast.

But that would be stupid because the whole point of the test was to see if the OCP on those PSUs is set to a consistent and reasonable point (no) and whether it does the job of saving the PSU from catastrophic failure (no).

I'm not so good at this - just to clarify here then: The 'core' problem here is that OCP is set way too high, and that if OCP was tighter, these might be fine units? Or is there also other component problems present?

SourKraut
Nov 20, 2005

CAUTION: POST QUALITY UNDER CONSTRUCTION





Canine Blues Arooo posted:

I'm not so good at this - just to clarify here then: The 'core' problem here is that OCP is set way too high, and that if OCP was tighter, these might be fine units? Or is there also other component problems present?

There are almost certainly other problems; the OCP being set at 140% is probably just exacerbating the problem.

Warmachine
Jan 30, 2012





Klyith posted:

So I wanna follow up with one clarification about this, because I think Steve's kinda tossed-off comment has the potential to be misunderstood. (It's the one thing I didn't like in that vid, but he's not working from script so it's hard to say things in ways that are both comprehensible to the audience and fully accurate.)

Mosfets are turning on and off all the time; that's how they work. The thing that produces 'wear' on them in a PSU is current and heat*. The main reason that a fluctuating load produces more wear in a normal scenario is that the transient peaks are much worse for them than the valleys are good. I am sure that a PSU held at 50% load will last longer than one that's bouncing between 10% and 90% while averaging 50%. However, I'm equally sure that holding it steady at 90% will kill it much faster than either.

*most of the time the current & heat are the same thing, because more current = more heat. but when they're different, in i dunno a LN2-cooled PSU, over current still kills.


So TBQH driving the mosfets at >130% of their rated load is close to the worst thing you can do to them, which is why that example was dead after 2 minutes of torture. If their testing procedure involved twiddling the buttons on their load tester to simulate a real-world load, it probably wouldn't kill them as fast.

But that would be stupid because the whole point of the test was to see if the OCP on those PSUs is set to a consistent and reasonable point (no) and whether it does the job of saving the PSU from catastrophic failure (no).



(I'm writing :words: just to make sure that nobody gets ideas like "mining bitcoin will make my PC last longer because steady state load is better than fluctuating!")

I stopped JUST short of saying "heat" in my post, since I'm so used to immediately translating a work load into a heat load. Literally thinking "MOSFET get hot, MOSFET get cold." And the only reason I don't shrug it off now is that I write stuff for a living where I have to go into tedious specifics like that because I can't guarantee the reader will understand the distinction when it is important.

Canine Blues Arooo posted:

I'm not so good at this - just to clarify here then: The 'core' problem here is that OCP is set way too high, and that if OCP was tighter, these might be fine units? Or is there also other component problems present?

No, the core problem is "something else." OCP failing to stop a catastrophic failure is a failure of your OCP implementation. While you presumably inform your choice of OCP by the rating of the hardware it is protecting, the actual trip point is arbitrary. High trip points aren't great because failures are a probability curve, and the more out of spec you get for longer, the higher the chance of a catastrophic failure.

The interesting insight is what they said about timings, which I interpret to mean polling rates. If the OCP isn't checking the status of the current draw often enough, it won't catch damaging transients and overcurrent situations fast enough to prevent damage (remember, time is a factor in the failure probability) or destruction.

In my non-electrical engineer lots of internet guy mind, either the parts downstream of the OCP are getting damaged before OCP trips because it doesn't trip fast enough or the OCP itself is faulty and is causing the catastrophic failure state. This is supported by how the PSU detonates at 60% load after the second OCP trip. Assuming QA tripped it once to make sure it was working, that means after 3 OCP trips, the components (namely that big rear end capacitor) were damaged enough to explode.

Someone wanna check me on that?

AlternateAccount
Apr 25, 2005
FYGM

64bit_Dophins posted:

Really? Is there some kind of 'me too' sort of stuff going on here?

I've never been able to watch the guy because he looks like a sludge monster and has the posture of a turtle, but he certainly looks like the kind of guy that would be a sex offender registry so nothing would really surprise me.

Thatís some pretty aggressive extrapolation. You can just say heís a nerd with questionable hardware restoration/analysis ďpracticesĒ and some broader political opinions you donít dig and a bizarre need to performatively get weird in Subways.

I mean I find his taste in sandwiches at least as alarming as his need for a rifle in there.

Warmachine
Jan 30, 2012





AlternateAccount posted:

Thatís some pretty aggressive extrapolation. You can just say heís a nerd with questionable hardware restoration/analysis ďpracticesĒ and some broader political opinions you donít dig and a bizarre need to performatively get weird in Subways.

I mean I find his taste in sandwiches at least as alarming as his need for a rifle in there.

Don't shame a man for enjoying elastic bread.

Klyith
Aug 3, 2007

GBS Pledge Week


Canine Blues Arooo posted:

I'm not so good at this - just to clarify here then: The 'core' problem here is that OCP is set way too high, and that if OCP was tighter, these might be fine units? Or is there also other component problems present?

Definitely the problem goes beyond the OPP being too high, and it may very well be that OPP too high is the secondary problem! Like, potentially the other stuff is even worse and the OPP just makes the failures fast and spectacular.

Steve and Partick mention in the video that the other reviewer Aris, who I guess does freelance articles for major tech sites about PSUs as well as runs a independent engineering certification company, thinks the driver frequency for the mosfets is too high.

And then the "other contact in the industry" is pointing to maybe a "thermal management issue". My immediate suspicion for who that contact they can't name might be is Jon Gerow, who used to run the jonnygugu PSU review site (:rip:) and now is the head of Corsair's PSU R&D. He seems to have a good relationship with GN. On Q&A videos when there's a question about general PSU details and Steve starts by saying "I didn't know so I reached out to...", frequently one of the replies is Jon.


So two actual electrical engineers who know their poo poo say it's hosed, but have different hunches for why it's hosed. Steve & Patrick aren't engineers but put tons of effort testing them under various conditions -- not all their failures were during / directly after the OPP test -- to show a >50% failure rate with many unsafe failures.

In conclusion, it's extremely hosed and nobody who owns one should feel ok using it just because they have a lower-spec system that won't ever draw high watts.



Warmachine posted:

The interesting insight is what they said about timings, which I interpret to mean polling rates.
Someone wanna check me on that?

No, they didn't say the full correct term in that vid (again, no script). The thing that Aris is talking about as a likely problem is the mosfet driving frequency -- that's what I mentioned before about power MOSFETs not being continuously on, they're pulsed on/off. How fast you pulse them is the driver frequency.

This is where I'm rapidly leaving my knowledge because I'm not an EE, and the friend who knows more is on vacation this week so all I've got is google. I do know that the faster you switch mosfets, the less efficient they are.

FuturePastNow
May 19, 2014




One of the models in question was reviewed by TechPowerUp last December- and it exploded on them. Their article has some good photos of the inside of the thing:

https://www.techpowerup.com/review/gigabyte-gp-p750gm-750-w/

Their conclusion after tearing it down and examining it is that the design is clean and the build quality is high- but the key components that failed were rock-bottom crap with some of the capacitors not even identifying their manufacturer. And that component selection is on Gigabyte.

Combat Pretzel
Jun 23, 2004

No, seriously... what kurds?!

Klyith posted:

I do know that the faster you switch mosfets, the less efficient they are.
Less efficient means more waste heat. More waste heat means things go kaplowy rather sooner than later.

FuturePastNow posted:

Their conclusion after tearing it down and examining it is that the design is clean and the build quality is high- but the key components that failed were rock-bottom crap with some of the capacitors not even identifying their manufacturer. And that component selection is on Gigabyte.
That's why I'm avoiding all their poo poo. Years back they already did the thing where they released the first one or two batches of product with decent components, making sure that the rat tail of reviewers didn't get any bad units, and then dropped component quality. And it seems like they're continuing to do so.

Combat Pretzel fucked around with this message at 09:31 on Aug 26, 2021

Dr. Video Games 0031
Jul 16, 2004



Their monitors are decent at least. They tend to make smart choices to reduce cost without sacrificing functionality, making for some pretty solid mid-budget displays that are competitive with the high-end displays out there. As for everything else they make, I think I'd rather stay away unless they're selling me a graphics card at msrp.

Speaking of gigabyte monitors, good news:

gradenko_2000
Oct 5, 2010

Ask me about mapping out all the best limousine routes in Moscow for you and the little miss ;)

Lipstick Apathy

lol at the techtuber community just passing around the same in-joke

der8auer did a review of an Alienware prebuilt and they just told people to refer to GN's breakdown of the Dell RTX 3090

Klyith
Aug 3, 2007

GBS Pledge Week


FuturePastNow posted:

One of the models in question was reviewed by TechPowerUp last December- and it exploded on them. Their article has some good photos of the inside of the thing:

https://www.techpowerup.com/review/gigabyte-gp-p750gm-750-w/

Their conclusion after tearing it down and examining it is that the design is clean and the build quality is high- but the key components that failed were rock-bottom crap with some of the capacitors not even identifying their manufacturer. And that component selection is on Gigabyte.

Check the byline, that's the review by Aris that Steve talks about.

By "design is clean" and "build quality is high" he's talking about the macro physical layout & soldering quality. Not the fundamental electrical circuit design. 'Cause like, if you can build a power supply that meets all the tests for voltage stability, ripple, gold efficiency, etc then surely you're not gently caress up the basic design of the thing, right?


Combat Pretzel posted:

Less efficient means more waste heat. More waste heat means things go kaplowy rather sooner than later.

Yeah but there are definitely other, more complicated factors that I don't understand. Internal capacitance that "must be charged or discharged", I'm guessing that driving it faster than it can reset that capacitance is a bad thing.

Anyways GN had a text note in the speedrun video that they had thermistors on stuff when they were doing the tests for real, and didn't see heat problems. OTOH when mosfets go into thermal runaway they get hot really fast, so an external sensor is useless. :shrug:

Adbot
ADBOT LOVES YOU

mA
Jul 9, 2001
I am the ugly lover.

Dr. Video Games 0031 posted:

Their monitors are decent at least. They tend to make smart choices to reduce cost without sacrificing functionality, making for some pretty solid mid-budget displays that are competitive with the high-end displays out there. As for everything else they make, I think I'd rather stay away unless they're selling me a graphics card at msrp.

Speaking of gigabyte monitors, good news:


They can be a decent economical choice (their mobos seem to be solid), but you're pretty hosed if your purchase has problems down the line even within the warranty period. If Gigabyte was my only choice, I'd consider buying a 3rd party warranty just so I don't have to deal with their clown car customer service or RMA experience.

mA fucked around with this message at 18:13 on Aug 26, 2021

  • 1
  • 2
  • 3
  • 4
  • 5
  • Post
  • Reply