GeForce GTX460 rumoured to launch at Computex

[happy medium]

Quote:

" The GTX460 is said to end up with 384 Shaders, that's 64 less than the GTX470 and 96 less than the GTX480. It is also said to feature 48 TMU's and 32 ROPs which in both cases is a reduction of eight compared to the GTX470. The GTX470 only got four TMU's and eight ROP's cut compared to the GTX480, so the GTX460 looks set to be a significantly slower part in relation to the GTX470."


Story here..............


http://www.semiaccurate.com/2010/04/23/geforce-gtx-460-rumoured-launch-computex/

 

[Lonyo]

Question is, will the cutbacks impact it like the HD5830, or will it be a more reasonable card?
And if it is clocked higher [presumably than a GTX470] then I wonder what that might do for power consumption.

ATI managed to (for the likes of us) royally screw up their gap-filling card by making something that no one really wants to buy.
NV hopefully won't go as wrong with their equivalent product. And maybe it might put some pressure on the HD5850 and HD5830 prices.


(And Computex is the first week of June, if anyone wasn't aware)

 

[happy medium]

Question is, will the cutbacks impact it like the HD5830, or will it be a more reasonable card?
And if it is clocked higher [presumably than a GTX470] then I wonder what that might do for power consumption.

ATI managed to (for the likes of us) royally screw up their gap-filling card by making something that no one really wants to buy.
NV hopefully won't go as wrong with their equivalent product. And maybe it might put some pressure on the HD5850 and HD5830 prices.

Been looking for a card about as fast as a gtx 285 with direct x 11. The 5830 was going to be the card, but that card is shit. I hope this brings the 5830 down to 190$,then mabe I'll bite.

I'll be interested in this gtx 460 if it's a little faster then a gtx 285 for under 250$.

 

[Alex@Nosound]

Two of theses in SLI and away i go

if there around GTX285 performance and not a total flop.

 

[GodisanAtheist]

Been looking for a card about as fast as a gtx 285 with direct x 11. The 5830 was going to be the card, but that card is shit. I hope this brings the 5830 down to 190$,then mabe I'll bite.

I'll be interested in this gtx 460 if it's a little faster then a gtx 285 for under 250$.

- Amen to that. I've lowered my expectations (on both sides of the fence) to the point that I'd just be happy for top o' th' line last gen performance for a cool $200 bucks. That's my sweet spot now, instead of hoping to double my performance (5850 n up) at that same price point.

 

[Attic]

Could shape up to be a bigdeal card if it shakes up the performance expectations at ~ $200.

 

[happy medium]

Two of theses in SLI and away i go

if there around GTX285 performance and not a total flop.

yea that sounds nice!
Should be about 5970 performance.

Should be about 200watts per card though.

 

[bunnyfubbles]

Question is, will the cutbacks impact it like the HD5830, or will it be a more reasonable card?
And if it is clocked higher [presumably than a GTX470] then I wonder what that might do for power consumption.

ATI managed to (for the likes of us) royally screw up their gap-filling card by making something that no one really wants to buy.
NV hopefully won't go as wrong with their equivalent product. And maybe it might put some pressure on the HD5850 and HD5830 prices.


(And Computex is the first week of June, if anyone wasn't aware)

if its based off the GF100 and not its own custom chip then its power consumption won't be great. Just like the 5830 vs. 5850, if a GTX460 is clocked higher than the 470 then it will likely consume more power and perform less. Although they'll save some power by using less memory.

Also, if we consider how much the 470 dropped off in performance vs. the 480, I'm not too hopeful about this card.

I'm more interested in an aggressively clocked GF104 part.

 

[toyota]

the only reason its cut down 64sp instead of just 32sp is that it will likely run at 600/1200 like the 470. if they had made it a 416sp part then they would have dropped the clocks to something really low like 500/1000. these might make for a nice sli setup that should smoke the gtx480.

 

[happy medium]

the only reason its cut down 64sp instead of just 32sp is that it will likely run at 600/1200 like the 470. if they had made it a 416sp part then they would have dropped the clocks to something really low like 500/1000. these might make for a nice sli setup that should smoke the gtx480.

2 gtx 460's would have 760 shaders for under 500$. Thats sounds nice.

 

[RaistlinZ]

2 gtx 460's would have 760 shaders for under 500$. Thats sounds nice.

A $250 MSRP would be nice, but I'd expect them to sell for $280 due to demad/price gauging. At that price I'd rather just go with dual 5850's, and I'm betting they'll still use less power than dual GTX 460's.

 

[happy medium]

A $250 MSRP would be nice, but I'd expect them to sell for $280 due to demad/price gauging. At that price I'd rather just go with dual 5850's, and I'm betting they'll still use less power than dual GTX 460's.

Does it rain everyday in your world?

 

[bunnyfubbles]

A $250 MSRP would be nice, but I'd expect them to sell for $280 due to demad/price gauging. At that price I'd rather just go with dual 5850's, and I'm betting they'll still use less power than dual GTX 460's.

If the GTX460 is a cut down GF100 and has the same clocks as a GTX470, it will likely have power characteristics similar to the 470, so that means more power than a Radeon 5870 let alone a 5850.

 

[badb0y]

Hoping it will be smack dab in middle of 5850 and 5770 but more closer to the 5850 side unlike the shitty ass 5830. Also no more than 200$ PLEASE!

 

[toyota]

Hoping it will be smack dab in middle of 5850 and 5770 but more closer to the 5850 side unlike the shitty ass 5830. Also no more than 200$ PLEASE!

around $250 sounds more realistic.

 

[Alex@Nosound]

Hoping it will be smack dab in middle of 5850 and 5770 but more closer to the 5850 side unlike the shitty ass 5830. Also no more than 200$ PLEASE!

Well GTX470 is £300 here so it would have to be like £220-250 (5850) pricing or it would ended up sort of like the HD5830 which has no market space :\

 

[Lonyo]

Well GTX470 is £300 here so it would have to be like £220-250 (5850) pricing or it would ended up sort of like the HD5830 which has no market space :\

The HD5830 has a giant market space.
The cheapest HD5770 1GB is £116, the cheapest HD5850 is £220.
That's a large space, and the HD5830 comes in at £190 (cheapest) which goes some way towards filling the gap.

A GTX460 priced equally with HD5850's in the UK wouldn't surprise me in the slightest though, even though it will most likely perform worse than the HD5850.
NV doesn't seem to like UK consumers very much.

 

[RaistlinZ]

Does it rain everyday in your world?

Based on what we've seen recently I have no reason to believe otherwise.

 

[MrK6]

For a competitive product, the HD5830 doesn't really make sense at USD prices. That said, it might be what it is for several reasons. My guess is the design is incredibly cheap to produce and therefore has a high profit margin, and provides a sink for bad 58xx chips (and there may not be many). That said, from a price/performance perspective, it should be <$200 MSRP. If the GTX 460 is competitive with the HD5830, it might take over that gap. I've always been curious as to why AMD didn't make a more aggressive attempt at that segment, maybe market data shows it really isn't profitable.

 

[A554SS1N]

If the GTX460 is a cut down GF100 and has the same clocks as a GTX470, it will likely have power characteristics similar to the 470, so that means more power than a Radeon 5870 let alone a 5850.

I put the figures into an excel spreadsheet, and if everything followed up the same, then the 460 should have 85% of the power consumption of a 470, putting it just slightly less than a 5870. Which is obviously still not good, as it'll probably perform around 5830 territory.

 

[happy medium]

I put the figures into an excel spreadsheet, and if everything followed up the same, then the 460 should have 85% of the power consumption of a 470, putting it just slightly less than a 5870. Which is obviously still not good, as it'll probably perform around 5830 territory.

Sorry if I'm oversimplifying this ,but if a gtx 285 with 240 sp's destroys a 5830 how would a gtx 460 , 384 sp part fail vs the 5830?

I'm guessing performance between a gtx 285 and 5850. The 5830 is about 9% faster then a 5770 , but slower then a 4890.

That just dosen't make sense to me.

 

[Lonyo]

Sorry if I'm oversimplifying this ,but if a gtx 285 with 240 sp's destroys a 5830 how would a gtx 460 , 384 sp part fail vs the 5830?

I'm guessing performance between a gtx 285 and 5850. The 5830 is about 9% faster then a 5770 , but slower then a 4890.

That just dosen't make sense to me.

It doesn't make much sense using just those figures that an HD5830 would be only 9% faster than an HD5770 and slower than an HD4890 (1120 vs 800 vs 800 SPs).

There's no real way to predict how it will end up performing, since we can't be entirely sure either what the specs are, or how the cutbacks will impact performance, so it's going to be a waiting game really.

It could be that with low clocks and due to the bits which have been reduced it will lose more performance than it "should", and would end up worse than a GTX285, or it could be very good.

 

[GodisanAtheist]

Sorry if I'm oversimplifying this ,but if a gtx 285 with 240 sp's destroys a 5830 how would a gtx 460 , 384 sp part fail vs the 5830?

I'm guessing performance between a gtx 285 and 5850. The 5830 is about 9% faster then a 5770 , but slower then a 4890.

That just dosen't make sense to me.

-Erm... for the same reason the 5830 has 1120SPs but is still slower than the 4890 with 800 SPs? For the same reason a GTX 470 has 90% more SPs than a GTX 285 but only 20% more performance? For the same reason the GTX 280 had 90% more SPs than the 9800GTX but was only 30% faster?

Because the SPs and their clusters are not perfectly comparable between this and the last gen. GPGPU modifications, Clustering, schedulers and the like can create a certain amount of overhead that might not have been there before, and perhaps a slight rearching of the individual SP itself could contribute to the difference in performance (get the SP to use fewer transistors/more transistors thanks to the 40 nm boondogle or who knows).

 

[Madcatatlas]

This card, even if it were as good as happy medium wants it to be, is just not for me. Im holding out for a singel high end card, which i can then Crossfire/Sli.

Thing is, i want to do this without even having to worry about heat, noise or power consumption. The 5870 would have been that deal if i was in the market 6 months ago.

Im looking forward to whatever comes next to be honest. Either a refresh of Fermi, or a 68xx

If these cards, in SLI, can surpass a gtx480 for 200 bucks each, it would seem a good deal.

 

[MrK6]

Sorry if I'm oversimplifying this ,but if a gtx 285 with 240 sp's destroys a 5830 how would a gtx 460 , 384 sp part fail vs the 5830?

I'm guessing performance between a gtx 285 and 5850. The 5830 is about 9% faster then a 5770 , but slower then a 4890.

That just dosen't make sense to me.

It probably wouldn't and shouldn't. The GTX 470 is about 20-25% faster than a GTX285 (if someone wants to refine that estimate, be my guest), but has almost double the shaders; performance isn't going to scale proportionally with shaders. However, if the specs given are true, the GTX 460 should be a bit faster than the GTX 285 without cannibalizing the GTX 470. An MSRP of $200-220 would sell like hotcakes, but then I would have thought a $300-320 MSRP for the GTX470 would have made more sense too. That said, I'd guess $250 MSRP and slightly faster than a GTX 285. Who knows, we could be pleasantly (or unpleasantly) surprised.