Overclock.net › Forums › Industry News › Rumors and Unconfirmed Articles › [Bit-Tech] Nvidia accused of crippling board partners' designs
New Posts  All Forums:Forum Nav:

[Bit-Tech] Nvidia accused of crippling board partners' designs - Page 16

post #151 of 252
Quote:
Originally Posted by duox View Post

For one amd would need to push out their next series while this is a hot topic and market online pointing it out ALOT. It is possible because amd does have a solid maketing and pr team, most my non enthusiast friends think amd is the best for gaming (0rocessors and graphics) simply because they have better advertising than intel and nvidia on the gaming front.
The 256 bit memory bus and annoying boost system have been a topic when discussing top of the line graphics cards for a while, so this is probably going to stay around for a while. It would be nice if Nvidia just dropped this BS because of all of the bad press, though. Nvidia had a great chip this generation, it was just limited by that stupid stock BIOS, the lack of good overclocking boards, and the 256 bit bus. Why did they never make a 685 with a 384 bit bus and charge $600 for it? That probably would have sold well.
post #152 of 252
Maybe Maxwell looks troubling for Nvidia early on, and just maybe "It's not what they are saying, but what they are not saying". Having great engineering headroom in the 600 series was too damn good. IS too damn good, and bad for Nvidia this fall and early next year. This hampers Next years line and forces them to engineer well beyond what they "Planned" to morsel out to the public.

Could you imagine if they released a GTX780 Jan 2013, and it was in the same performance range as some of the partner jacked up cards. Make no mistake they don't want to do a Bulldozer and fail to "Clearly" beat the last generation. It would be the kiss of death.

It appears like Nvidia was not happy they had to bring "all the the goods" with the 680 as AMD forced them to actually bring a good product out... Now they want to backpedal.

Nvidia Exec..."Lock it down" cigar.gif Besides it's worked for Intel...Many of us that had planned SR-X builds decided to stay home due to the Locked CPU's...my money went into an NSX. biggrin.gif


O.T. - No the car wont get you women. I get asked all the time by guys - "You get women with that car?...No, that's shallow, but to answer the question it's just the opposite. Every gearhead for ten miles comes walkin up...askin questions...sigh
post #153 of 252
Sweet NSX...
post #154 of 252
I perfectly understand what you mean. However I have a suspicion that instead of offering an unlocked low voltage with possibility of OV, nvidia engineers changed the approach and did some extensive testing on OV impact on card longevity (and RMAs related to this) ... it appears to me that they locked the max voltage at value based on their lab testing of many chips to rule out some insanely high % of RMAs in near future. Different approach it is, not everybody has to like it. I wouldn't be surprised if AMD follows the same path in near future instead of taking their chances to "win" more enthusiasts.

While this "safe" and warrantied approach is logical it still gives some crazy boost to "stock" performance as you noticed, that's a big thing. I believe the OV (high enough) over that max voltage value, set by nvidia, would cause a massive degradation of gpu on certain % of units over a short time (only nvidia knows the exact numbers) and they simply refuse to allow it for reference models (nvidia is liable for warranty replacements on those).
Vendors still can make their own unlocked models, they just need to cover replacements on their own, and I'm sure they will re-release them after some time, but with new warranty rules ... and pricing adjusted up for those who want a manual OV over the specs limits.

I used to own 570 SLI and loved those cards a lot, upgraded only for future proofing, meaning I preferred a single more powerful PCIe3.0 card on my new z77 build, so I could go SLI in future.
There was nothing wrong with 570 for me, used them for over a year, setup was stable and powerful at raised voltage (up to 1.13v when benchmarking) ... but to be honest 1.13V was still a software controlled voltage, the maximum I could set in Precision X. Also keep in mind that some 570 reference units came with lower or higher stock voltage - one of my cards was stock @ 0.963v while other was same stock @ 1.025v. Now with 600 series you get all cards operating in exactly same stock voltage range, so all of them are more equal, but still can overclock differently via manually controlled offsets, e.g. my 670 @ stock clocks runs 1097MHz core and with manual offset it runs at 1243MHz core - that is within the same safe/locked voltage limits.

So the approach of past was different, lower (varying between batches) and fixed factory voltage and some headroom for manual voltage tweaking up to max safe 1.13v (or 1.15v as I recall in MSI:AB or other software) while now you get a product which already allows that range of dynamic voltage adjustments in stock BIOS and does the clock boost automatically when needed, a different approach. With 600 series user can still manually control the Power Target and offsets on the core & memory clocks (that makes the card run even faster within the very same locked voltage range versus pure "stock").
Quote:
Originally Posted by Brutuz View Post

Why can't nVidia offer a low, but still unlocked voltage? And offer more insane voltages on top of that? That's my point, there's no good for the consumer reason not to do that...Only making it cheaper for nVidia which is arguably worse for consumers as we then have to deal with cheap VRMs and all their associated drawbacks (For example, increased heat generation even at stock voltages and a shorter lifespan), I somehow doubt we'd get the savings passed onto us either..Also, keep in mind that the 570s only failed because of a bad batch of VRMs in an early model, later models were fine, the 590 had a few problems because..well, nVidia cheaped out on VRMs; for that reason I doubt you'll see many functioning 590s in 5 years from now, especially compared to the amount of 580s and 570s, that's not even taking the amount sold into account, just that the 590s VRM was minimum for stock operation.
Why can I pick between HD7970 and GTX 680 MSI Lightnings, ASUS DirectCUs and Gigabyte SuperOverclocks if manufacturers can't be on both sides? That was plain pigheadedness from nVidia.
He owns a site and hates nVidia, but has some great insiders and great pre-release information if you ignore his fanboy problems.
You want to know the main reason I think this is bad for everyone, not just us? Competition. Remember, nVidia has GPU Boost to get OCed numbers against AMDs stock cards (Not that it's a bad thing, in fact it's great for laptops and consumers) so this will directly impact how high GPU Boost can let the cards clock if Nvidia goes cheap on VRMs due to this, it will also affect longevity among other things; I won't be boycotting nVidia simply because of this but if AMDs cards OC higher due to having OV and also run faster due to that...I'll definitely be going to AMD. However, I have a feeling this is in the best interests for nVidia and that knowing the GPU industry nVidia will be back to software overvolts within 5 generations of cards.

Edited by feniks - 10/6/12 at 1:55pm
post #155 of 252
Quote:
Originally Posted by feniks View Post

I perfectly understand what you mean. However I have a suspicion that instead of offering an unlocked low voltage with possibility of OV, nvidia engineers changed the approach and did some extensive testing on OV impact on card longevity (and RMAs related to this) ... it appears to me that they locked the max voltage at value based on their lab testing of many chips to rule out some insanely high % of RMAs in near future. Different approach it is, not everybody has to like it. I wouldn't be surprised if AMD follows the same path in near future instead of taking their chances to "win" more enthusiasts.
Flashing the BIOS voids the warranty. If the cards started to degrade and lose stability over time, then it could be possible to flash the BIOS back and RMA it. How many people actually do that, though?
Quote:
While this "safe" and warrantied approach is logical it still gives some crazy boost to "stock" performance as you noticed, that's a big thing. I believe the OV (high enough) over that max voltage value, set by nvidia, would cause a massive degradation of gpu on certain % of units over a short time (only nvidia knows the exact numbers) and they simply refuse to allow it for reference models (nvidia is liable for warranty replacements on those).
Vendors still can make their own unlocked models, they just need to cover replacements on their own, and I'm sure they will re-release them after some time, but with new warranty rules ... and pricing adjusted up for those who want a manual OV over the specs limits.
The problem with that is that Nvidia gives no warranty on any of the cards that have the possibility of overvolting, so the cards being DOA, and Nvidia is probably selling all of the cards for a profit, so they are making money off selling defective cards as well. I doubt businesses will ever be able to do this.
Quote:
I used to own 570 SLI and loved those cards a lot, upgraded only for future proofing, meaning I preferred a single more powerful PCIe3.0 card on my new z77 build, so I could go SLI in future.
There was nothing wrong with 570 for me, used them for over a year, setup was stable and powerful at raised voltage (up to 1.13v when benchmarking) ... but to be honest 1.13V was still a software controlled voltage, the maximum I could set in Precision X. Also keep in mind that some 570 reference units came with lower or higher stock voltage - one of my cards was stock @ 0.963v while other was same stock @ 1.025v. Now with 600 series you get all cards operating in exactly same stock voltage range, so all of them are more equal, but still can overclock differently via manually controlled offsets, e.g. my 670 @ stock clocks runs 1097MHz core and with manual offset it runs at 1243MHz core - that is within the same safe/locked voltage limits.
So the approach of past was different, lower (variable) fixed factory voltage and some headroom for manual voltage tweaking up to max safe 1.13v while now you get a product which already allows that range of dynamic voltage adjustments in stock BIOS and does the clock boost automatically when needed, a different approach. With 600 series user can still manually control the Power Target and offsets on the core & memory clocks (that makes the card run even faster within the very same locked voltage range versus pure "stock").
I don't want to sound like a dick, but upgrading for the sole purpose of future proofing is pretty pointless, and kind of defeats the point of future proofing.
post #156 of 252
I never said I had to flash the BIOS to achieve my max boost clocks I quoted on 670! it's on stock BIOS... 1243MHz core (+148MHz versus stock boost without core offsets) with warranty from vendor (and nvidia warranty to vendor). With flashing BIOS (voiding warranty) you could get even higher clocks (ca. +40-60MHz higher on top of max I quoted before).

Nvidia gives no warranty TO VENDOR, but vendor CAN give the warranty TO YOU IF they have the balls to do that with their custom OV designs smile.gif other story that most of them got scared right now and are re-calculating the bottom lines if that was to happen.

It's not up to you to challenge or judge my future proofing approach, sorry, that sentence of yours was pointless and very subjective. I want to have 2x relatively inexpensive PCIe3.0 cards in system in near future, that is why I upgraded to Z77+IB, why? because I like it. will I upgrade video cards in 2 years to newer models? yes, probably I will while keeping the same board+cpu.
Quote:
Originally Posted by Art Vanelay View Post

Flashing the BIOS voids the warranty. If the cards started to degrade and lose stability over time, then it could be possible to flash the BIOS back and RMA it. How many people actually do that, though?
The problem with that is that Nvidia gives no warranty on any of the cards that have the possibility of overvolting, so the cards being DOA, and Nvidia is probably selling all of the cards for a profit, so they are making money off selling defective cards as well. I doubt businesses will ever be able to do this.
I don't want to sound like a dick, but upgrading for the sole purpose of future proofing is pretty pointless, and kind of defeats the point of future proofing.

Edited by feniks - 10/6/12 at 2:41pm
post #157 of 252
Quote:
Originally Posted by feniks View Post

I never said I had to flash the BIOS to achieve my max boost clocks I quoted on 670!
I was responding to this comment:
Quote:
they locked the max voltage at value based on their lab testing of many chips to rule out some insanely high % of RMAs in near future
Quote:
Nvidia gives no warranty TO VENDOR, but vendor CAN give the warranty TO YOU IF they have the balls to do that with their custom OV designs smile.gif other story that most of them got scared right now and are re-calculating the bottom lines if that was to happen.
but without the warranty they can no longer return cards that had manufacturing defects, so they would probably lose a lot of money on them, and this will prevent any manufacturers to economically release overvolting graphics cards.
Quote:
It's not up to you to challenge or judge my future proofing approach, sorry, that sentence of yours was pointless and very subjective. I want to have 2x relatively inexpensive PCIe3.0 cards in system in near future, that is why I upgraded to Z77+IB, why? because I like it. will I upgrade video cards in 2 years to newer models? yes, probably I will while keeping the same board+cpu.
If you're planning on upgrading in the near future, then it's not really future proofing, it's just upgrading. If you're upgrading just to extend the period of time that you will have before you need to upgrade (futureproofing), that doesn't make much sense.
post #158 of 252
and why are you worried about the vendor being able or unable to get money for a broken card which allowed OV way above nvidia specs and was custom modded by that vendor? it's their problem, neither mine nor yours. of course it would probably touch the users who NEED to buy cards allowing for custom OV, but at same time there is much bigger risk of burning such cards and the vendor needs to offset such cost of potentially higher RMA in their price (since nvidia gives them a finger on it). I was actually surprised that nvidia allowed for such a long time highly modified cards with designs totally way above their specs to get replaced by vendors for free (or parts of it) under warranty. I guess nvidia grew tired for paying for failures of other non-reference designs, and personally I'd do the same, just way sooner biggrin.gif
vendors need to take responsibility for modifications they make out of specs instead of counting on the good uncle paying for their mistakes, simple as that.

obviously my understanding and approach to future proofing is way different from yours, fine with me. I like to replace my video cards for new models at least every 2 years and re-sell them while they are still worth something (and preferably when still in warranty), so I upgrade them to newer/stronger models that use less power for same performance. My electric bill matters to me to some extent.
By future proofing I mostly mean keeping the board+cpu for 3-4 years, I don't stick to old video cards when I find them vastly too power hungry for same performance as newer model, I skip certain series though, my former cards before 500 series were 200 series. 600 series just does the trick for me, period. probably I will skip 700 series and wait for the next one for the upgrade.
Quote:
Originally Posted by Art Vanelay View Post

but without the warranty they can no longer return cards that had manufacturing defects, so they would probably lose a lot of money on them, and this will prevent any manufacturers to economically release overvolting graphics cards.
If you're planning on upgrading in the near future, then it's not really future proofing, it's just upgrading. If you're upgrading just to extend the period of time that you will have before you need to upgrade (futureproofing), that doesn't make much sense.

Edited by feniks - 10/6/12 at 3:48pm
post #159 of 252
Quote:
Originally Posted by feniks View Post

and why are you worried about the vendor being able or unable to get money for a broken card which allowed OV way above nvidia specs and was custom modded by that vendor? it's their problem, neither mine nor yours. of course it would probably touch the users who NEED to buy cards allowing for custom OV, but at same time there is much bigger risk of burning such cards and the vendor needs to offset such cost of potentially higher RMA in their price (since nvidia gives them a finger on it). I was actually surprised that nvidia allowed for such a long time highly modified cards with designs totally way above their specs to get replaced by vendors for free (or parts of it) under warranty. I guess nvidia grew tired for paying for failures of other non-reference designs, and personally I'd do the same, just way sooner biggrin.gif
vendors need to take responsibility for modifications they make out of specs instead of counting on the good uncle paying for their mistakes, simple as that.
obviously my understanding and approach to future proofing is way different from yours, fine with me. I like to replace my video cards for new models at least every 2 years and re-sell them while they are still worth something (and preferably when still in warranty), so I upgrade them to newer/stronger models that use less power for same performance. My electric bill matters to me to some extent.
By future proofing I mostly mean keeping the board+cpu for 3-4 years, I don't stick to old video cards when I find them vastly too power hungry for same performance as newer model, I skip certain series though, my former cards before 500 series were 200 series. 600 series just does the trick for me, period. probably I will skip 700 series and wait for the next one for the upgrade.
How many times do I have to say that flashing the BIOS voids the warranty? If the voltage is limited in the BIOS, then people will have to void their warranty to increase the voltage to levels which will seriously affect longevity,
post #160 of 252
Quote:
Originally Posted by Art Vanelay View Post

How many times do I have to say that flashing the BIOS voids the warranty? If the voltage is limited in the BIOS, then people will have to void their warranty to increase the voltage to levels which will seriously affect longevity,

LOL! dude, you are weird ... my point was that you do not have to flash the BIOS since even with vmodded BIOS the gains are small and only around +50MHz over the max you can do with STOCK FREAKING BIOS AND MANUAL OFFSETS AND MAX POWER TARGET WITHIN MANUFACTURER SPECS WHILE KEEPING THE WARRANTY! You are still subject to lottery when buying those cards and nobody knows if you buy a high clocker or a mediocre one, this never changes ever.

let me put this straight forward for you (570 reference unlocked voltage vs 670 reference locked voltage):
1. stock 570 @ 732MHz and whatever vgpu factory set for it (could be 0.96v or could be 1.025v)
2. max stable overclocked 570 @ around 900-950MHz with max allowed vgpu up to 1.13v (based on experience with two such cards)
3. stock 670 doing max STOCK boost of 1097MHz with stock 100% power target value and automatic/adaptive voltage within specs (goes up to 1.17V when needed, idles at 324Mhz @ 0.9v in adaptive power mode)
4. max stable overclocked 670 doing max MANUAL MODE OFFSET on the core with MAX power target value (within specs) resulting in max boost at 1243MHz core and automatic/adaptive voltage within specs (still goes up to stock 1.17v when needed and idles at 324Mhz @ 0.9v in adaptive power mode)

comparison results:
570 reference can on average get overclocked (within safe specs) by around 168MHz (for my weaker card) or up to 218MHz (for my stronger card) when going with max allowed safe voltage
670 reference can on average get overclocked (within safe specs) by around 146MHz when going with max allowed safe voltage (locked by nvidia)
was warranty voided for either of those cards? no.

would you void the warranty on 570 card when pumping it with much higher voltage than overclocking software allowed 1.13v (or 1.15v for some apps)? yes, I would think so. same as you would by running 670 card with 1.21v voltage (vmodded BIOS), nobody is forcing you to do so however and choice is always upon user (with consequences).

performance of each is another story and in short a single 670 gives nearly same gaming experience as 570 running in SLI, benchmarking scores (synthetic) will of course be higher for 570 still, but so will the power usage and generated heat.

if you meant classified/lightning cards with your comment then write to vendors who pulled them down and ask when/if they are going to release the new custom OV models with their own warranty (since nvidia will not pay for RMA) covering possible failures coming from their custom designs utilizing nvidia components.

I am done with this thread, do as you will people, run to AMD if you like, not my problem, really ... subscribing out.
Edited by feniks - 10/7/12 at 6:20pm
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: Rumors and Unconfirmed Articles
Overclock.net › Forums › Industry News › Rumors and Unconfirmed Articles › [Bit-Tech] Nvidia accused of crippling board partners' designs