Read more.The first retail Maxwell card examined.
Read more.The first retail Maxwell card examined.
Do the fans on these shut off completely if not gaming?
Wasabi, nope.
Pay more than 15% over reference for a 5% performance boost and a comically large cooler. Bargain.
"Nothing is safer than a giant snowball whipping through space...at a million miles an hour"
Yup. You'd have to be a fool to buy this Asus model at £135; as you say, a R9 270 would serve it a beat-down at a lower price, as would the GTX 660 whilst availability remains. And there's no advantage to the Asus 750Ti over cheaper, more sensibly cooled, versions.
750Ti is far and away the best card you can get which doesn't need an auxiliary power connector (AMD's best is the 7750), and has a TDP low enough that it should fit on a low profile bracket with a single slot cooler (if a 7750 can do it, so can these) - which would be perfect for very-SFF builds. No one has made one though.
Technically speaking the fastest bus powered card ever made was the slot powered HD7850 2GB which was available in the Far East for OEMs. However,it appeard to be an expensive to make design.
Originally Posted by Toms Hardware
The non-deterministic boost means the card is frequently breaking the PCI-E power limit for very short periods,and even then the HD7750 consumes less power than a GTX750TI and relies on no boost at all.
TPU measures the HD7750 under Crysis2 as consuming 43W to 45W. The GTX750TI consumes between 52W to 69W plus it relies on boosting beyond TDP spec for short periods.
It needs a better cooler than the HD7750 for these reasons alone. Maybe not a massive dual fan heatsink,but hardly a tiny slab of aluminium,as it would just throttle probably after a extended gaming section in a cramped low profile case. Most reviews are done on open air benches or full ATX cases.
It is probably why we have not seen any singe slot low profile GTX750TI cards ATM.
Last edited by CAT-THE-FIFTH; 20-02-2014 at 12:57 PM.
In Crysis 3, Anandtech found a total system draw of 10w difference. Given that the 750Ti outperforms the 7750 by well over 50% in Crysis 3 (per Anandtech's review), not all of that 10w will be the GPU (CPU working harder)
It's worth remembering that single slot versions of the 7850 exist (as do low-profile ones, though not low-profile AND single slot), and that's a significantly hotter GPU.
http://www.kitguru.net/components/gr...ssfire-review/
The variation between the normal HD7850 and GTX750TI is around 18W to 35W according to TPU who measure card power consumption at the connectors. However,the card I was talking about bus powered,so was probably using binned chips:
http://www.techpowerup.com/185003/tr...n-hd-7850.html
However,remember the HD7850 does use a 43% larger die(it uses the same chip as an HD7870) so the GPU will run cooler. The HD7850 also uses a 256 bit memory controller and this means more complex PCBs with more RAM chips,which raises power consumed,so not all of that increase is attributable to the GPU.
But it was hardly available,so yes the GTX750TI is the best easily available slot powered card.
However,power measurements are at the PCI-E slot for TPU too,so yes the GTX750TI does consume more power than a HD7750,anything up 26W more at the slot.
The TH review also measured card power consumption. The card goes past spec when boosting. The HD7750 does not boost and is lower power. Hence it can get away with worse cooling.
Nvidia Boost 2.0 has problems over extended periods with less than optimal cooling. Hardware.fr,pcgameshardware and HT4U showed throttling problems with the Geforce Titan and GTX760 models using reference coolers over extended periods,and it reduced performance a noticeable amount.
Whereas a massive twin fan heatsink might be OTT,a tiny heatsink in a small case will cause problems,when the card is boosting.
If you read reviews carefully,the first version of Nvidia Boost had problems(which I kind of predicted might occur during the GK104 launch with aggressive non-deterministic boost tech),and in the end many review sites had to pre-heat the cards before running benchmarks to maintain consistency.
The first iteration of the AMD boosting tech was deterministic and the boost was tiny,so had less problems. The moment they went with more aggressive boosting in the R9 290 series the same issues happened,although it is deterministic and has a maximum cap.
All this boosting tech does,is make GPU cooling a more important factor now. CPUs tend to not be anywhere as aggressive as CPUs when it comes to boosting,and are far more conservative. This is why there have been far less issues with them having boost clockspeeds.
All it seems to do is boost AMD and Nvidia benchmark scores in the short term.
Edit!!
Look at the Nvidia mobile cards in the GTX600 and GTX700 which have no boost unlike their desktop counterparts:
http://en.wikipedia.org/wiki/GeForce_600_Series
http://en.wikipedia.org/wiki/GeForce_700_Series
All are fixed clockspeed designs.
That indicates they probably know you need reasonable cooling for it to work.
Last edited by CAT-THE-FIFTH; 20-02-2014 at 01:16 PM.
Wish this cooler worked this well on R9 290.
Anyone got any decent stable overclocks on this above the stock OC?
There are currently 1 users browsing this thread. (0 members and 1 guests)