ASUS GeForce GTX 750 Ti "Maxwell" Sees First Listing

[btarunr]

Way ahead of its rumored February 18 launch, a Hungarian online retailer listed an ASUS-branded GeForce GTX 750 Ti graphics card, bearing model number GTX750TI-OC-2GD5. The card is listed at a tax-inclusive price of 50,000 HUF (around US $225), which is rather steep, even for a pre-order, given the positioning of the card, in relation to existing products, such as the GTX 660 and GTX 760. The naming suggests that the card is factory-overclocked, which means that NVIDIA will allow factory-overclocked, and possibly custom design GTX 750 Ti cards from day-one. It also gives away the memory of the card, at 2 GB GDDR5. The listing, however, doesn't go with any pictures.



View at TechPowerUp Main Site

 

[leeb2013]

it's a bit steep, although if it matches or beats a 760 (around $260) it's not too bad.

 

[15th Warlock]

So the first Maxwell comes out in February... and a rather anemic release at that... oh well, I hope the 20nm parts are not far behind.

To be honest I'm really interested in seeing what Maxwell will bring in terms of power efficiency to mobile platforms, we will see how Nvidia play their cards in that arena.

 

[OC-Rage]

Good nvidia im waiting FOR 750Ti Good price

 

[MxPhenom 216]

it's a bit steep, although if it matches or beats a 760 (around $260) it's not too bad.

Rumors are saying it is 15% slower than the GTX660.

 

[LeonVolcove]

any specs about this thing?

 

[ZetZet]

If it performs like 650ti boost and costs the same it would be pointless, so I think it should be close to 760.

 

[Asagrim]

The VAT in Hungary is 27%, which is included in the price, so take that in consideration before you compare the price with other cards sold elsewhere in USD.

 

[Casecutter]

it's a bit steep, although if it matches or beats a 760 (around $260) it's not too bad.

Ah, why would Nvidia (or anyone) consider under-cutting their own current product segment?

The VAT in Hungary is 27%, which is included in the price, so take that in consideration before you compare the price with other cards sold elsewhere in USD.

Good point and from that we might say it maybe something like $165-170 USD, or less if there some pre-order upcharge in that price.

But IDK... I'd consider Nvidia would like to get right into the sub $200 and displace the R9 270/270X using a die that's give Nvidia higher profits than a GK106 appears to be able to via.

 

[HumanSmoke]

Ah, why would Nvidia (or anyone) consider under-cutting their own current product segment?

You mean like the GTX 770 undercutting the GTX 680 ?
Or the GTX 760 undercutting the similarly performing GTX 670?
Or the GTX 780 undercutting Titan ?

 

[Casecutter]

Ah, why would Nvidia (or anyone) consider under-cutting their own current product segment?

But to displace a recently released segment (well just shy of 6 mo.) that carries the present or up-to-date naming convention with a lower designation (and lower price) would be a little strange. The only reason I might see that happening is if the GK104 isn’t holding to the profit schedule they originally targeted and they don’t want to go to war (price drops) with such a card.

 

[carowden]

You mean like the GTX 770 undercutting the GTX 680 ?
Or the GTX 760 undercutting the similarly performing GTX 670?
Or the GTX 780 undercutting Titan ?

all of the ones you are referring to (except for the 780, im assuming you meant "780 ti", and the titan) are examples of a new iteration of cards taking the place of an older part. thats not was he was saying. he was saying it wouldnt make sense to release a 750ti that undercut the CURRENT generation 760 in terms of price AND performance. even with the 780ti and titan, its not really undercutting, it just pushed the titan into a different segment as more of a productivity card and solidified the 780ti as a gaming card.

 

[HumanSmoke]

all of the ones you are referring to (except for the 780, im assuming you meant "780 ti", and the titan) are examples of a new iteration of cards taking the place of an older part.

Which is only a valid argument if the 600 series cards went EOL when the new parts launched. That was not the case. 670 and 680 cards sold alongside the newer 700 cards for some time....something Casecutter himself bought up at the time

he was saying it wouldnt make sense to release a 750ti that undercut the CURRENT generation 760 in terms of price AND performance.

In general terms it would be unlikely that Nvidia would indeed launch a lower numbered 750 that beats out a 760, but it isn't totally unheard of. The small die GTX 460 undercut the large die GTX 465 in exactly the same manner. Likely the Maxwell die is smaller than GK 104, no? The GTX 275 both outperformed and was significantly cheaper than the GTX 280. There is also the option for Nvidia to prioritize the GPU for a narrower feature set, which might elevate its raw frames per second performance, but be prejudicial in a larger context (lower/no FP64 ability for example).
Personally it is a moot point since the earlier leak clearly stated that the card is positioned between the 650 Ti and 660, but one thing is certain in graphics cards - never rule anything out until the product actually launches.

 

[alwayssts]

This is a chip that has to satisfy essentially 2 criteria:

1. Be smaller than gk106/Pitcairn.
2. Be faster than Bonaire (260x) and 7850...one would think even situated to be viable against the inevitable 20nm shrink of Bonaire (So, essentially faster than amd with 896sp at ~1100/7000 on a 128-bit bus at stock) and vicariously worthy of competition with the next amd apu after Kaveri which is prolly also 896sp.

But it also seems highly unlikely it will attempt to challenge the 270 series.
It would also make sense it could be shrank to 20nm, which implies 128-bit bus.

That doesn't leave a whole lot of options besides something with a fairly similar unit construction (while maxwell may be different in sp/sfu smx structure to kepler) to Bonaire/650ti boost in a native config.

Perhaps something like 4*240sp (doing away with the 192sp+32sfu of kepler) smx at 1037/7000?

That would make sense from a power consumption level (1.05v should be able to obtain that clock) and also increase their gpgpu/compute performance by 25% per clock over 650ti (960 vs 768). Not to mention, if their memory bus stays intact and could scale somewhere in the vacinity of 7600-8000, the core could scale to 1125-1185mhz, or around the edge of the power curve for 28nm (if not 20nm)...ie ~1.163-1.175-1.2v. One would think it would have to be under 150w max tdp, so also take that into consideration.

That would be a rather efficient and forward-thinking design...although still not beat a 270 in any conceivable manner. It would however likely be cheap to produce (outside of memory) with good yields and relatively good power consumption (again, outside of the memory; high clocks consume a lot of power relatively). While it would not reign supreme at under 150w (because of 270 vanilla), it would be faster than Bonaire, perhaps only slightly larger (hence more lucrative than 270 is for amd, if not allowing it to be priced significantly cheaper), and probably make them a fortune through volume sales to people that want something just a little faster than said chip or amd's next apu.