• Competitor rules

    Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.

EVGA 460 2WIN

Are you sure the memory still works like that? How do you know it is not 2GB shared ram. Surely, it won't run like SLI, but like a dual GPU card?

I would want more than 1GB
Dragon Age 2 @ 1920x1080 was using 1.2GB ram on my 2gb gtx 460!

I'm looking for another 2GB for SLI, so been reading about the memory in SLI a lot.
 
You're right seekdestroy. If they had just lobbed two GPUs together and SLI'd em on the PCB then it would be 1GB effective. BUT that's not to say they couldn't have redesigned it so that both processors use a single shared 2GB buffer. Significantly harder to achieve, but something like that would be epic. It will be a lot better than 2 GTX 460s in SLI.

I read on Bi-tech or somewhere they they plan to price it below a GTX 580. :/ that makes me sad. I was hoping for GTx 570 like prices. But I guess even priced less than 580 people will buy it because it gives better performance. And 460 scaling is as near-ideal as it gets. Power consumption is also lower than GTX 580. Add to that overclockability. I doubt this will run hot as some have claimed. My DirectCU II 460 overclocked to 931.5 Mhz/1863 Mhz/4200 at 1.1volt never tops 52C in a game. this card looks mega long. i am guessing its still got some of that 460 overclockability.

Now let's tap our feet impatiently while we wait for GTX 560 2WIN
 
seekdestroy said:
It would be great to find some more info, and benchmarks. It's 11.5 inches long! and weighs 3lb
Click to expand...

Indeed. The EVGA benchmarks don't show enough because they're at Performance mode 3D Mark 11 and 1440x900 Heaven. Both low resolutions. It will be good to see how the performance scales from 1080p to 1600p. If perf at 1600p/1200p with high AA/AF starts to fall off rapidly then we'll know it's a frame buffer limitation.
 
bhavv said:
I went from MSI GTX 460s to 560s, and had removed the coolers on each to reapply the paste, both PCBs were identical and the same size as far as I could tell. If it was any longer, it wouldnt fit into may case without removing a fan.
Click to expand...

Ok, MSI might have made their own PCB, they usually do, then overspec it for better overclocking, in other words a over specced 460gtx was more than capable of running a normal spec 560gtx, but it still stands that the 460/560gtx pcb's aren't the same, neither are the power requirements, look up Anandtech's reviews of both, they always show the pcb, its longer and has more substantial power circuitry.

xsistor said:
Agree about the memory. They should have taken GTX 460 768MB cards and doubled the memory/ROPs to 1536 MB/48 and then paired two of these.

That would've given us a Dual GTX 460 3GB - but with an effective frame buffer of 1536 MB. Plenty sufficient for the type of thing you'd use a 460 SLI for.


But even this will probably work great. I mean honestly the people using this 460 2WIN will be those gaming at 1080p or 1200p and for that 1GB is enough
Click to expand...

You can't just double the rop's, its a design, its a core, it has X amount of rops available. the 1gb version has 32 rops, thats all you're going to get out of it. Adding more rops would be a 6 month design turnaround and require taping out and a complete redesign of the core. You probably can put 1.5gb mem on each core, by using double density ram, when the 58xx/4xx cards launched you could only get 128mb chips, you can get 256mb ones now and is how AMD doubled memory capacity with the same bus.


seekdestroy said:
Are you sure the memory still works like that? How do you know it is not 2GB shared ram. Surely, it won't run like SLI, but like a dual GPU card?

I would want more than 1GB
Dragon Age 2 @ 1920x1080 was using 1.2GB ram on my 2gb gtx 460!

I'm looking for another 2GB for SLI, so been reading about the memory in SLI a lot.
Click to expand...

Its 2gb total, 1gb is usable by each core, its effectively a 1gb card, ALL dual gpu and xfire/sli setups work like this.

If that changed it wouldn't be on a year old architecture it would be HUGE news on a new architecture.
 
drunkenmaster said:
You can't just double the rop's, its a design, its a core, it has X amount of rops available. the 1gb version has 32 rops, thats all you're going to get out of it. Adding more rops would be a 6 month design turnaround and require taping out and a complete redesign of the core. You probably can put 1.5gb mem on each core, by using double density ram, when the 58xx/4xx cards launched you could only get 128mb chips, you can get 256mb ones now and is how AMD doubled memory capacity with the same bus.
Click to expand...

I think you need to read my previous post. I said they should've redesigned it. It would've added more circuitry and been more extensive but something like that would've made it more worthwhile than just SLIing two 460s to end up with a card that might not work well at high resolutions.
And no not by redoing the chip. 24 rops per chip. 48 rops total, but rework how the memory is interfaced to the chip so that they share the same 1.5GB memory instead of just going 1GB/1GB for 1GB total.
 
Last edited:
xsistor said:
Agree about the memory. They should have taken GTX 460 768MB cards and doubled the memory/ROPs to 1536 MB/48 and then paired two of these.

That would've given us a Dual GTX 460 3GB - but with an effective frame buffer of 1536 MB. Plenty sufficient for the type of thing you'd use a 460 SLI for.
Click to expand...

Yeah, I need to reread it, more like you need to talk sense.

Firstly the card as is, the 460 2win has 64 Rops already, 32 per core, so why you'd recommend using smaller cores with less rops on a more expensive and powerful card, well, makes no sense.

As for a redesign, you should try reading what I wrote. You can double the capacity but using double density memory chips with NO OTHER DESIGN CHANGE.

The only thing that makes a 460gtx 768 is a 1gb version with a marginally cut down bus(which drops memory chip number) and cutting off a few rops.

You still are saying they could just do a small redesign, a small redesign involves making a new chip, even with minor changes would take 3-4 weeks to design correctly and several months to tape out and test.

You realise they can, both use 256mb chips and make it a 4gb card, or have cut the bus marginally, kept the rops and had 1.5gb per core, for which there would be very little reason.

The cost difference, between 2gb and 1.5gb, would be two whole 256mb chips per core, or, about $15 for the entire card, but to drop to 1.5gb you'd be killing 25% of the bandwidth, 25% less bandwidth, on a card faster than a 580gtx, to save $15, or with your idea, save $15, kill bandwidth dramatically and randomly reduce the amount of rops also by 25%..............

Oh, and just reworking the memory interface(instead of using higher capacity chips) would require reworking the memory interface, which is part of the gpu, which would require a full new mask, tape out and months to produce, test, validate, all to be, slower than the card they have already.


http://www.anandtech.com/show/3835/gainward-announces-a-2gb-gtx-460

2gb 460gtx's that version launched in September, doubling memory capacity by using double density chips is OLD news, like really old, and why you'd use two inferior cut back cores and 1.5gb per core instead of 2gb, which has already been done loads of times, I don't know. If you want a card that will do great at high res, why go lower on mem than you can, and why cut out 25% of the performance for realistically no reason, and yes, at uber high resolutions bandwidth becomes every bit as important as actual amount.
 
Last edited:
ryan4227 said:
Hey

What happens with drivers for this thing. It's not using an Nvidia reference design or Gpu config?

Regards Ryan
Click to expand...

Nothing really, it would use bog standard drivers, to put it very simply the chip between the two cores on an AMD/Nvidia dual gpu card is really just a pci-e interface chip so the driver simply see's 2 separate cards, connected by a pci-e bus, which is all it see's if you took two separate cards and put them in different slots.

The only difference with dual gpu cards(for several generations and including upcoming ones it would seem) is convienience of two cards on one pcb, theres little to no functional difference, except differing clock speeds if required to keep temps managable.

Theres really 2 reasons to use one, you have a silly case like a HTPC that doesn't have enough pci-e slots or space for x-fire, 98% of these cases and setups don't want to be noisy and can't take a card that long anyway so its almost pointless. The other lot are guys who will go quadfire/sli, have more money than sense(in most situations, if you're doing gpu intensive work and make money from your work, its more than worth it) and fitting 4 cards in any case/mobo is a pain in the behind, and 2x dual gpu cards has advantages, theres almost never another good reason to use them.

AMD in the past offered great value which was really the only reason to go for them, the 3870x2, 4870x2, and the rrp at least of the 5970 were significantly lower than the 2 equivilant separate cards, the downsides were still not worth it though.
 
You can only have one of these cards anyway so unless its cheaper then two theres no point buying it.
EVGA have 10% off when u buy two gtx460's ssc.
 
drunkenmaster said:
Yeah, I need to reread it, more like you need to talk sense.

Firstly the card as is, the 460 2win has 64 Rops already, 32 per core, so why you'd recommend using smaller cores with less rops on a more expensive and powerful card, well, makes no sense.

As for a redesign, you should try reading what I wrote. You can double the capacity but using double density memory chips with NO OTHER DESIGN CHANGE.

The only thing that makes a 460gtx 768 is a 1gb version with a marginally cut down bus(which drops memory chip number) and cutting off a few rops.

You still are saying they could just do a small redesign, a small redesign involves making a new chip, even with minor changes would take 3-4 weeks to design correctly and several months to tape out and test.

You realise they can, both use 256mb chips and make it a 4gb card, or have cut the bus marginally, kept the rops and had 1.5gb per core, for which there would be very little reason.

The cost difference, between 2gb and 1.5gb, would be two whole 256mb chips per core, or, about $15 for the entire card, but to drop to 1.5gb you'd be killing 25% of the bandwidth, 25% less bandwidth, on a card faster than a 580gtx, to save $15, or with your idea, save $15, kill bandwidth dramatically and randomly reduce the amount of rops also by 25%..............

Oh, and just reworking the memory interface(instead of using higher capacity chips) would require reworking the memory interface, which is part of the gpu, which would require a full new mask, tape out and months to produce, test, validate, all to be, slower than the card they have already.


http://www.anandtech.com/show/3835/gainward-announces-a-2gb-gtx-460

2gb 460gtx's that version launched in September, doubling memory capacity by using double density chips is OLD news, like really old, and why you'd use two inferior cut back cores and 1.5gb per core instead of 2gb, which has already been done loads of times, I don't know. If you want a card that will do great at high res, why go lower on mem than you can, and why cut out 25% of the performance for realistically no reason, and yes, at uber high resolutions bandwidth becomes every bit as important as actual amount.
Click to expand...

And you need to stop getting your panties in a bunch.

Because SeekandDestroy suggested that it should be possible to design multi-GPU cards in such a way that they use the same memory. And I Was talking about that. Redoing the chip is one way. ROPs are decoupled in GPU pipelines these days and so, yes it is possible. But that's not what I was suggesting. If you thought it was then I'll apologise for my poor way of communicating it to you. Rather, I was suggesting going with the 768MB cards because if they're doing a shared memory version of it that is all that's really necessary for a twin 460 card to be great.

Remaking the chips is usually a bad idea because of the high initial cost of ASICs. Since this is entirely a digital circuit it is doable on FPGAs instead for small volumes, but that would still make it pricier than existing chips that are ASIC. I'm speaking from experience. Some years ago I designed an experimental processor, for real-time image processing and for nonlinear robust control, that had to go on ASIC instead of FPGA because even though volume was small it was a mixed signal circuit. that's was a different issue entirely because I work with analogue computation which is far far more powerful than digital. and this was a mixed-signal circuit done in VHDL with CMOS floor plans for the digital aspect, so only ASIC would do.

What I was talking about instead is using two 460s with 768MB 24 ROPs each for a 48 ROP card with shared 1.5GB memory, which would really be ideal for virtually anyone who wants a dual GPU card in that price range. It is highly likely you can do that with external control circuitry without modifying the chips themselves. Of cousre it is impossible to tell precisely without looking at the card design itself whether it would be feasible, but to me it looks like it should be possible. Nothing ground breaking needs to be done here either. I can see several ways they may set about doing soemthing like that.


If the whole point of your rant was to say that such a card is not economically feasible, then noted. I cannot agree or disagree unless I'd actually studied the designs and looked at how costly it would be, but I have a notion there is a fairly simple solution to doing his in practice. If it can't be done fairly easily then it makes sense to just stick to hte 1GB cards. or go with 2GB cards and have 4GB. But I often in engineering for something like this where you're dealing with essentially the same data on two buffers there are simpler ways of achieving this. For example, connect circuitry that fools one GPU into thinking that its fetch requests are being satisfied, and then feed it data from the same memory buffer when it makes a request. Again, impossible to say if that will work without actually looking at the designs.

But unlike you I am not going to knock an idea based on some superficial information I read off random review sites who are written mostly by people who aren't even microelectronics engineers themselves.
 
Last edited:
xsistor said:
What I was talking about instead is using two 460s with 768MB 24 ROPs each for a 48 ROP card with shared 1.5GB memory, which would really be ideal for virtually anyone who wants a dual GPU card in that price range.
Click to expand...

That may compete too much with the 480. remember that they still have quite a few they still want to get rid of
 
Just got to wait for the dual gpu goodness of the 590 to come out now. If it uses a cooler similar to that it will be a beast :)
 
drunkenmaster said:
AMD in the past offered great value which was really the only reason to go for them, the 3870x2, 4870x2, and the rrp at least of the 5970 were significantly lower than the 2 equivilant separate cards, the downsides were still not worth it though.
Click to expand...

True, value was the clincher with these cards for most of the time, and AMD has usually delivered. I wonder what has made them depart from that to such a degree with the 6990? It is far far too expensive at the moment, especially considering you can spend £400 on two 6950s which will at least equal if not beat it!
 
You must log in or register to reply here.
Back
Top Bottom