[vr-zone] GTX 590 revision in June

[OCGuy]

I don't get it? What didn't "work right the first time" for the GTX590?

For some reason when the 590 was launched, it became the end-user's right to overvolt graphics cards without risk of failure.

 

[SolMiester]

What I find funny, is most of the comments are all negative instead of positive for a better product...
A lot of negative people around I guess, kinda feel sorry for them!

 

[AdamK47]

At this point I'm more concerend with any news on next gen nVidia GPUs.

 

[Kenmitch]

What I find funny, is most of the comments are all negative instead of positive for a better product...
A lot of negative people around I guess, kinda feel sorry for them!

I guess at first it does look negative....But I see it as pun!

I'd like to see some photos of the pcb's to see what's changed.

 

[Firestorm007]

I don't get it? What didn't "work right the first time" for the GTX590?

I don't see this situation as being any different than CPU steppings, or mobo revisions. Unless the stepping change is to fix something uber-bad (like the B3 stepping for Phenom that fixed the TLB bug), stepping changes bring performance improvements on terms of overclockability all the time for all camps.

Q6600 B3 -> G0?

PhII C2 -> E0?

OC'ing is always YMMV, and new steppings generally bring improved overclocking. This is true of motherboards with new revisions as well.

I agree. However, it looks like there was some sort of hardware respin on these. For some reason, they've been OOS at Newegg for some time now. Maybe they were silently pulled for the fix. I can't be sure. IDC, I don't pretend to know more than you on these things. You've got me beat there. However, there was something fundamentally wrong with the initial PCB design on this card. If some overvolting resulted in these things burning up, there's a problem. It may seem like I'm indulging in this, but I'm not. I own both an EVGA 580GTX (Great Card by the way) and a few AMD solutions (5870/5770). What I find odd is that some people pretend that there wasn't some initial problem with these.

 

[Tempered81]

I guess at first it does look negative....But I see it as pun!

I'd like to see some photos of the pcb's to see what's changed.

I'll bet you the new PCB has the same power connectors, same gpus, same memory. But it has more inductors, more caps, more power phases, more slaves, and robust controller, and that all the VRM components are higher quality parts.

I think they should call it the 595

 

[blastingcap]

Wake me up when 28nm arrives... and it arrives as soon as this fall. These monster dual-40nm cards will be hot, noisy energy hogs that likely can't beat the next generation top-end parts. I don't know why people would bother unless they need that kind of horsepower RIGHT NOW and/or have money to burn.

 

[Grooveriding]

Wake me up when 28nm arrives... and it arrives as soon as this fall. These monster dual-40nm cards will be hot, noisy energy hogs that likely can't beat the next generation top-end parts. I don't know why people would bother unless they need that kind of horsepower RIGHT NOW and/or have money to burn.

I agree. The 590 had its chance and failed, this late fix to the flawed PCB is not going to do anything to change the landscape.

I have my $ ready for two 7970s hopefully in September, and if nvidia can manage to have cards ready with a release date announced within two weeks of the 7970 launch date, maybe for two 680s if they are faster.

 

[Spikesoldier]

I'll bet you the new PCB has the same power connectors, same gpus, same memory. But it has more inductors, more caps, more power phases, more slaves, and robust controller, and that all the VRM components are higher quality parts.

I think they should call it the 595

agreed. this is a little more intensive than just a core revision.

 

[Spikesoldier]

For some reason when the 590 was launched, it became the end-user's right to overvolt graphics cards without risk of failure.

LOL so why is it that in the case of the 590 that the failures can be reportedly be duplicated with ease?

some cards can handle an overvolt. obviously from a design perspective the 590 cannot. yes people run a risk when overvolting their card, but nothing like the death wish on the 590.

the 590 was rushed to market and thus suffered design issues or lack of QA in design process. people usually rush when they are late, and this accurately describes fermi and the 590's disposition.

had they spent a little extra time on the design process, im sure they could have worked the bugs out and delivered a quality and competitive product.

 

[mosox]

Good news: Nvidia will replace all the defective GTX 590s with brand new HD 4650 cards.

 

[wahdangun]

I don't get it? What didn't "work right the first time" for the GTX590?

I don't see this situation as being any different than CPU steppings, or mobo revisions. Unless the stepping change is to fix something uber-bad (like the B3 stepping for Phenom that fixed the TLB bug), stepping changes bring performance improvements on terms of overclockability all the time for all camps.

Q6600 B3 -> G0?

PhII C2 -> E0?

OC'ing is always YMMV, and new steppings generally bring improved overclocking. This is true of motherboards with new revisions as well.

yes. I know all of that idc, but its not that simple especially some of the AIB, advertising oc as one tools to make them look different and superior than other AIB,

 

[3DVagabond]

Wake me up when 28nm arrives... and it arrives as soon as this fall. These monster dual-40nm cards will be hot, noisy energy hogs that likely can't beat the next generation top-end parts. I don't know why people would bother unless they need that kind of horsepower RIGHT NOW and/or have money to burn.

Agree 100%! Having to stay at 40nm for these cards was a major fail.

agreed. this is a little more intensive than just a core revision.

A 580 or 6970 Lightning is still a 580 or 6970. Upgrading the power stages doesn't justify naming it a new card.

I don't get it? What didn't "work right the first time" for the GTX590?

I don't see this situation as being any different than CPU steppings, or mobo revisions. Unless the stepping change is to fix something uber-bad (like the B3 stepping for Phenom that fixed the TLB bug), stepping changes bring performance improvements on terms of overclockability all the time for all camps.

Q6600 B3 -> G0?

PhII C2 -> E0?

OC'ing is always YMMV, and new steppings generally bring improved overclocking. This is true of motherboards with new revisions as well.

Why are they coming out with a revision if all was OK? This is not a new stepping or respin. It's a redesign of the power stage because the original design is sub par.

They were pressured to match AMD and release a dual card. They already missed one entire gen with the 400 series. It would have been better for nVidia's customers if they had held off release until they had redesigned it to compete with the 6990. I don't think they had any idea that AMD was going to push the power envelope so far. They probably had already made them though and decided to push them as far as they would go in an attempt to match the 6990 and sell them. I don't think they are up to the standards of other nVidia cards. If I bought an original release design I'd be very disappointed if nVidia doesn't offer replacement with a new card.

 

[Arkadrel]

Wake me up when 28nm arrives... and it arrives as soon as this fall. These monster dual-40nm cards will be hot, noisy energy hogs that likely can't beat the next generation top-end parts. I don't know why people would bother unless they need that kind of horsepower RIGHT NOW and/or have money to burn.

+1 ^-^

I didnt even bother with the 6xxx series of the 5xx one, my 5870 is still pretty near the top and didnt consider it worth while to upg. Come 7xxx series, I ll likely upg again.

 

[Gikaseixas]

good news for future owners.

The GTX 590 is a great card for those who just want a fast video card and don't wanna mess with it. For the extreme user it's not so good due to the problems we all know so i think nVidia just beefed up the card for that purpose. They made a mistake but now at least they're making the right call.

6990 is still king but it will have a tougher job when the updated 590 arrives. Tests with overclocked versions will be interesting.

I'm looking forward to 28nm.

 

[Tempered81]

6990 > 590, but IMO, 590 rev2 > 6990. Once the board is robust enough to handle two full gf110's at high speeds, Nvidia will have the performance crown again.

 

[Silverforce11]

I agree. The 590 had its chance and failed, this late fix to the flawed PCB is not going to do anything to change the landscape.

I have my $ ready for two 7970s hopefully in September, and if nvidia can manage to have cards ready with a release date announced within two weeks of the 7970 launch date, maybe for two 680s if they are faster.

Same, i decided to skip 69xx and go to 28nm for my new rig. In time for BF3 on eyeinfinity! Here's hoping for a smooth 28nm product launch in Q3-4.

 

[Jionix]

I wouldn't be surprised to see a new cooler as well. If they are finally outfitting the card with the needed power components to run the card faster, it will need cooling to match.

And this isn't at all akin to new processor stepping. This is a hardware design change, not a manufacturing process improvement.

 

[Firestorm007]

+1 ^-^

I didnt even bother with the 6xxx series of the 5xx one, my 5870 is still pretty near the top and didnt consider it worth while to upg. Come 7xxx series, I ll likely upg again.

+1 -- IMHO the 6 series just didn't offer enough of a bump in performance for me. I was kind of disappointed. But, there was only so much that AMD could do with 40nm. I hope the 7 series will be great.

 

[Arkadrel]

I suspect alot of people will just wait out until 7xxx series is out, and then upg to be able to play Battlefield 3 smoothly with all the settings on high.

I know I wouldnt throw down money for a 590 or 6990 (even though those are out of my price class) at this point, simply because of how big the jump from 40nm->28nm is, and what thats bound to do for the cards.

 

[exar333]

Awesome, they will probably once again own the fastest card and single GPU.

Personally I am set until the 28nm launches in Q4.

This x1000!

I am still very happy with my 5870 and the current gen (NV: 5xxx and AMD 6xxx) brought very little to the table. Looking forward to a real new card on 28nm.

 

[bryanW1995]

I'm not a fanboy and I don't scream, but I thought that the original card was fine. When used in the way that it was designed, it worked. People pushed it out of the design envelope and it failed. I think it was more a case of enthusiast overclockers expecting too much because they have become spoiled with all the easy overclocks common to hardware these days.

What? The ultra high-end cards are highly likely to be pushed/tweaked/overclocked/etc. Nvidia pushed their clocks up too high to try for the performance title, and the card was underengineered. TPU and others have consisitently run the same/similar tests on every card released for many years, nvidia should have performed those same tests in-house before releasing the card to ensure that it actually worked. Even better, they should have spent the extra few bucks to build the card correctly in the first place.

Look at it this way, when was the last time that nvidia (or amd for that matter) had even a single card blow up during reviewer testing? Maybe the 5990 part uno will be a good soldier and work for several years at stock clocks, but reason and intuition tell me that anybody spending $700 + on a video card would be much more apt to get a 6990 atm. I really like my nvidia experience over the past several years, and in particular I like the different levels off AA and custom profiles, but if I was getting a dual gpu ultra high end card now it would definitely be from amd. Good call on nvidia to actually fix the problem at least, though I'm surprised that it took them so long.

I don't get it? What didn't "work right the first time" for the GTX590?

I don't see this situation as being any different than CPU steppings, or mobo revisions. Unless the stepping change is to fix something uber-bad (like the B3 stepping for Phenom that fixed the TLB bug), stepping changes bring performance improvements on terms of overclockability all the time for all camps.

Q6600 B3 -> G0?

PhII C2 -> E0?

OC'ing is always YMMV, and new steppings generally bring improved overclocking. This is true of motherboards with new revisions as well.

A cpu ananogy would say that gtx 590 is k8 and 6990 is conroe. Only if they were originally clocked with ~ the same performance.

 

[bryanW1995]

At this point I'm more concerend with any news on next gen nVidia GPUs.

What I find funny, is most of the comments are all negative instead of positive for a better product...
A lot of negative people around I guess, kinda feel sorry for them!

The response is so negative b/c the fix is not very timely. We're getting too close to the buildup in anticipation for 28nm; most people who wanted a ginormous sandwich already got one or settled for 2 x GTX 580 instead. If they'd immediately halted sales of gtx 590 and had this fix out 10 days later then NV would have gotten some deserved criticism but also some kudos for immediately addressing the problem. By waiting 2 months, they have clearly said to us one again that their high end strategy only involves dual-gpu cards when absolutely necessary. Personally, I don't like AMD's dual gpu strategy at all, I think that single gpus (perhaps in xfire/sli) is a far better solution, but if NV is going to do it then they should at least put enough resources into it to be competitive.

 

[Firestorm007]

The response is so negative b/c the fix is not very timely. We're getting too close to the buildup in anticipation for 28nm; most people who wanted a ginormous sandwich already got one or settled for 2 x GTX 580 instead. If they'd immediately halted sales of gtx 590 and had this fix out 10 days later then NV would have gotten some deserved criticism but also some kudos for immediately addressing the problem. By waiting 2 months, they have clearly said to us one again that their high end strategy only involves dual-gpu cards when absolutely necessary. Personally, I don't like AMD's dual gpu strategy at all, I think that single gpus (perhaps in xfire/sli) is a far better solution, but if NV is going to do it then they should at least put enough resources into it to be competitive.

Very well said!

 

[betasub]

most people who wanted a ginormous sandwich already got one or settled for 2 x GTX 580 instead.

The 590 has been called many things, but single PCB means not a sandwich: