Gtx 650 ti has more Tmu's than 750 ti

[mroofie]

Do you think its a downgrade ?
The 750 ti has 16 rops just like the 650 ti
But the 750 ti has less tmu's

 

[Chetkigaming]

It's like a steadygrade.

 

[Slizzo]

Do you think its a downgrade ?
The 750 ti has 16 rops just like the 650 ti
But the 750 ti has less tmu's

750Ti outperforms 650Ti in all tests I have seen. Progress is progress.

You can't compare GPU make up when there has been a GPU architecture redesign.

 

[Easy Rhino]

Do you think its a downgrade ?
The 750 ti has 16 rops just like the 650 ti
But the 750 ti has less tmu's

Question is: Why would anyone go from a 650Ti to a 750Ti ?

 

[Hilux SSRG]

Question is: Why would anyone go from a 650Ti to a 750Ti ?

Maybe they thought it was an upgrade? Many of my friends [less techy] see a higher number and think its way better.

 

[mroofie]

750Ti outperforms 650Ti in all tests I have seen. Progress is progress.

You can't compare GPU make up when there has been a GPU architecture redesign.

So the amount of Tmu's needed can be part of a architecture redesign ?

 

[mroofie]

Question is: Why would anyone go from a 650Ti to a 750Ti ?

I dont have a 650 ti ! im just pointing something out -__-

 

[erocker]

So the amount of Tmu's needed can be part of a architecture redesign ?

Yes.

 

[True Christian]

I had the same kind of question when i looked at the GTX 660. It had 'only' 24 ROPS compared to my old GTX 460 1GBs 32 ROPS but was 70% faster. I wasn't treated kindly by the trolls but I learned why.

 

[cadaveca]

So the amount of Tmu's needed can be part of a architecture redesign ?

TMU = Texture Memory Unit.

As such, the actual specifications (size and speed) of TMU can vary, specifically with a change in GPU design architecture.

Think of TMU like cache... it's not exactly the same, of course.

 

[Fluffmeister]

I dont have a 650 ti ! im just pointing something out -__-

Superb thread then either way.

I say go Maxwell, fast and efficient, but then so is Kepler.

People here are already upset about the possibility of some mid-range Maxwell part ruining AMD's high-end party.

Bless em.

 

[ArbitraryAffection]

The key to the 750 Ti's performance advantage lies in how it uses its resources; the Maxwell architecture can achieve far higher resource utilisation due to an architectural re-design, which is also why the 750 Ti is faster despite having less CUDA cores than its predecessor (640 vs 768). Also, as a 750 Ti owner; I'll vouch for how well they overclock, too. I've got mine running at 1413 / 6.6 and it's rock solid.

 

[Slizzo]

The key to the 750 Ti's performance advantage lies in how it uses its resources; the Maxwell architecture can achieve far higher resource utilisation due to an architectural re-design, which is also why the 750 Ti is faster despite having less CUDA cores than its predecessor (640 vs 768). Also, as a 750 Ti owner; I'll vouch for how well they overclock, too. I've got mine running at 1413 / 6.6 and it's rock solid.

Which is amazing really, as it's still on the 28nm process and not on the 20nm process like it should be.

 

[ArbitraryAffection]

Which is amazing really, as it's still on the 28nm process and not on the 20nm process like it should be.

Indeed it is, I'm very much looking forward to seeing the level of performance and efficiency Maxwell will bring on the 20nm process (if it goes ahead as planned). Though part of me believes the overclocking capability of the 750 series is likely due to the maturity of 28nm silicon.

 

[mroofie]

TMU = Texture Memory Unit.

As such, the actual specifications (size and speed) of TMU can vary, specifically with a change in GPU design architecture.

Think of TMU like cache... it's not exactly the same, of course.

Maxwell also has 2mb cache vs 1mb kepler
another improvement

 

[mroofie]

The key to the 750 Ti's performance advantage lies in how it uses its resources; the Maxwell architecture can achieve far higher resource utilisation due to an architectural re-design, which is also why the 750 Ti is faster despite having less CUDA cores than its predecessor (640 vs 768). Also, as a 750 Ti owner; I'll vouch for how well they overclock, too. I've got mine running at 1413 / 6.6 and it's rock solid.

Evga ?

 

[Hilux SSRG]

Also, as a 750 Ti owner; I'll vouch for how well they overclock, too. I've got mine running at 1413 / 6.6 and it's rock solid.

That's a pretty awesome oc for your card. The 1413 is base clock or boost?

 

[mroofie]

Indeed it is, I'm very much looking forward to seeing the level of performance and efficiency Maxwell will bring on the 20nm process (if it goes ahead as planned). Though part of me believes the overclocking capability of the 750 series is likely due to the maturity of 28nm silicon.

Which 750 ti was that ? please tell me

 

[mroofie]

The key to the 750 Ti's performance advantage lies in how it uses its resources; the Maxwell architecture can achieve far higher resource utilisation due to an architectural re-design, which is also why the 750 Ti is faster despite having less CUDA cores than its predecessor (640 vs 768). Also, as a 750 Ti owner; I'll vouch for how well they overclock, too. I've got mine running at 1413 / 6.6 and it's rock solid.

jo please tell me what manufacturer you are using please tell me !!!! :'(

 

[ArbitraryAffection]

That's a pretty awesome oc for your card. The 1413 is base clock or boost?

The boost clock reported under load (using MSI Afterburner).

Evga ?

Yup. EVGA SuperClocked, doesn't even need a power connector.

 

[TheBrainyOne]

I have no idea what this means.

Whereas the SMX was for all practical purposes a large, flat design with 4 warp schedulers and 15 different execution blocks, the SMM has been heavily partitioned. Physically each SMM is still one contiguous unit, not really all that different from an SMX. But logically the execution blocks which each warp scheduler can access have been greatly curtailed.

The end result is that in an SMX the 4 warp schedulers would share most of their execution resources and work out which warp was on which execution resource for any given cycle. But on an SMM, the warp schedulers are removed from each other and given complete dominion over a far smaller collection of execution resources. No longer do warp schedulers have to share FP32 CUDA cores, special function units, or load/store units, as each of those is replicated across each partition. Only texture units and FP64 CUDA cores are shared.
Among the changes NVIDIA made to reduce power consumption, this is among the greatest. Shared resources, though extremely useful when you have the workloads to fill them, do have drawbacks. They’re wasting space and power if not fed, the crossbar to connect all of them is not particularly cheap on a power or area basis, and there is additional scheduling overhead from having to coordinate the actions of those warp schedulers. By forgoing the shared resources NVIDIA loses out on some of the performance benefits from the design, but what they gain in power and space efficiency more than makes up for it.