Discussion PSU recommendations and power supply discussion thread - Tom's hardware

Page 45 - Seeking answers? Join the Tom's Hardware community: where nearly two million members share solutions and discuss the latest tech.

4745454b

Titan
Moderator
I'm not saying the N1 is a good or great unit. But I have an issue with that review because he's changing the scoring again. If you ignore the crossload results the scores turned in aren't that bad and in spec. 0A on the 3.3V and 5V rail? When is that going to happen? I know it can happen as that is why we are seeing 12V only PSUs coming out, but I still feel this is unrealistic. The other half of this issue is I doubt all the other reviews have been updated with the scoring change. It's possible there is another unit with even worse out of spec issues that got a pass because it was reviewed X months before the change happened. I love reading JG site and the reviews are usually rather good. But it's changes like this that ruin it for me.
 

Elisis

Commendable
Nov 14, 2019
40
4
1,545
from these 2 which you would buy??
Super Flower Leadex Titanium 1600W Power Supply or Corsair AX1600i ??
Out of those I would probably recommend the AX1600i, as it does have (configurable) multi rail OCP over the Leadex.
 
Last edited:
This actually leads me to a question... on older systems (pre-Haswell Intel, or AMD pre-Ryzen), these are absolutely fine, even with, say, a more modern video card?

I thought I'd read something about them not handling power demand changes well on modern systems, but if that's false, then, ugh, know I've erroneously passed that incorrect info on.

I am unaware of any information showing conclusively that there is any kind of problem running these units with pre-Haswell Intel platforms OR with post-Haswell platforms so long as the C6 and C7 low power states are disabled, or with ANY AMD platform, since none of the AMD platforms have a similar implementation to what causes that problem on the Intel platforms. If there is, I've either not seen it or I've seen it and it wasn't substantiated so I forgot about it.

I'm pretty sure if there WAS a problem, we'd have seen a MUCH bigger and continuing stink about it than anything that we've seen in the past when the whole Haswell low power state thing became a popular subject of discussion there for a while. In fact, I thought I'd read something recently saying that the problem was actually somewhat overblown ANYHOW, even on the newer Intel platforms, but that playing it safe was a good idea anyway because of the obviously inherent flaws in those group regulated platforms including lacking in efficiency, lacking in protections and lacking in some other features found on newer platform designs and implementations.

But "work"? I've seen nothing that suggests there is a fundamental reason why they can't be used with ANY platform, at all, so long as you take the appropriate steps to ensure that C6/C7 are avoided if it is a platform that will try to use it.

If there is compelling evidence saying these shouldn't be used with modern platforms for OTHER concerns, I'd sure like to see it.
 

King_V

Illustrious
Ambassador
I'm trying to go by memory, but I want to say it was with regard to how sudden demand spikes can occur with RX580/590 and Vega cards.

I think it was in a comment somewhere in some thread on this forum, but I have no idea who'd said it, or even if they were in a position to know what they were talking about, and there wasn't an explanation as to why/how it would be a problem.

(I've really got to remember to bookmark it when I come across bits like that....)
 
Those cards had problems with MANY power supplies. Rogue leader had those problems with a 550w Prime Ultra titanium, on a Vega card. So I don't think that problem is strictly related to the group regulated design. It might be something inherent in VARIOUS designs. I'd have to know more to know for sure but I don't recommend the group regulated models unless they are the only decent available option and if they are, then I do. Pretty simple actually.
 
  • Like
Reactions: King_V

bignastyid

Titan
Moderator
I'm trying to go by memory, but I want to say it was with regard to how sudden demand spikes can occur with RX580/590 and Vega cards.

I think it was in a comment somewhere in some thread on this forum, but I have no idea who'd said it, and there wasn't an explanation as to why/how it would be a problem, or even if they were in a position to know what they were talking about.

(I've really got to remember to bookmark it when I come across bits like that....)

I have the Antec HCG620m(same platform as the M12II) in a rig with a i7 3820(no c6/7 sleep states) and RX580(Red Devil). No issues as of yet.
 

Rogue Leader

It's a trap!
Moderator
I'm trying to go by memory, but I want to say it was with regard to how sudden demand spikes can occur with RX580/590 and Vega cards.

I think it was in a comment somewhere in some thread on this forum, but I have no idea who'd said it, or even if they were in a position to know what they were talking about, and there wasn't an explanation as to why/how it would be a problem.

(I've really got to remember to bookmark it when I come across bits like that....)

Just Vega, not RX 580/590.

I have a Vega 64 LC, in testing this GPU on my system (1800X not overclocked) was able to transient spike the whole system to 900w, which is insane. This caused my PSU, which should have been enough for it, to shut down on protection. The GPU will sustain normally in Turbo mode 330w, and spike measurably around 380w. If you run a Vega in Balanced mode it won't go over about 270w which is way more reasonable but obviously a bit slower. The GPU does respond to undervolting so if you want to put the time and effort in you may be able to get turbo level performance at less wattage, but it is not a simple thing to do.

Needless to say, if you're running a Vega DO NOT skimp on the PSU, I certainly wouldn't use an older group regulated one, and I wouldn't use anything lower than 750w if you want max performance.

Those cards had problems with MANY power supplies. Rogue leader had those problems with a 550w Prime Ultra titanium, on a Vega card. So I don't think that problem is strictly related to the group regulated design. It might be something inherent in VARIOUS designs. I'd have to know more to know for sure but I don't recommend the group regulated models unless they are the only decent available option and if they are, then I do. Pretty simple actually.

It was actually a 650w Prime Titanium, the first one released (although the Ultra Titanium is basically the same thing with some minor changes). I replaced it with a 750w Ultra unit and have had 0 problems since, in fact power consumption went down a touch on the 750.
 
  • Like
Reactions: King_V

King_V

Illustrious
Ambassador
Just Vega, not RX 580/590.

I have a Vega 64 LC, in testing this GPU on my system (1800X not overclocked) was able to transient spike the whole system to 900w, which is insane.

I literally stopped and said in my own head "I misread that. I absolutely MUST have misread that."

But, I guess I didn't. "Insane" might well be the understatement of the century on that point...
 

Rogue Leader

It's a trap!
Moderator
I literally stopped and said in my own head "I misread that. I absolutely MUST have misread that."

But, I guess I didn't. "Insane" might well be the understatement of the century on that point...

Interestingly enough I wasn't totally aware of it, because when I was hitting protection it didn't make sense. I was monitoring power usage and it was close to maxed out but fine, and then randomly it would shut off. Someone else pointed out in this thread to a test that was done showing these transient spikes that were tripping protection, that were fast enough to not be picked up by a normal power meter.

I like my Vega, it looks cool and does the job. However I would never recommend it to anyone.
 
Sad that a fairly new series doesn't meet the latest ATX specs, especially since there are better units in it's price range that almost certainly will. Might be a good unit to add to the list for markets where it is available and others might not be though.

It's also unfortunate that it is somewhat noisy. Hard to recommend noisy power supplies these days in the price ranges of those models. Since the 750w version of that unit runs about 105 dollars, there are clearly better options at or near the same price.
 
Ultra brand was Tiger direct's house brand for MANY years. They are all pretty much crap. When Tiger direct folded a few years back, that was mostly the end of them. They were commonly included in the garbage bundles offered by Tiger direct.

I don't know how many systems I've fixed, literally, fifty to seventy-five or more, by simply removing the Ultra power supply, throwing it away and putting something decent in its place.

Jon Gerow actually used to work for them, once upon a long time ago.


They are and have pretty much always been, on everybody's "do not use" list.
 

4745454b

Titan
Moderator
I believe they also own the patent on the modular connection for PSUs. I don't know if anyone else has found a way around that. It's odd that such a bad company owns the rights to something so many use.
 
They don't. Their claims have been rejected, three times.

https://www.legitreviews.com/us-patent-office-rejects-ultra-products-claim-for-modular-psus_10103

Their patent is ONLY for the modular board used on their X-connect line of power supplies apparently. I have no clear understanding of the details regarding the patent or the lawsuit, but I do know that they've been shot down each time.

Plus, if they have not maintained the required patent fees periodically, they may not have any legitimate patent remaining at all, besides which, the patent protection expires entirely in 2024 in any case.
 

TRENDING THREADS