I can only hope NVidia optimizes the drivers in a hurry.
Einstein got a huge increase in productivity on all the Nvidia cards whose owners opted to allow Beta test applications simply by using the CUDA55 code generation path (as opposed to CUDA32). That has zero awareness of even Maxwell opportunities, and of course is only provided to users aware and interested enough to select the non default option. Clearly getting more out of the Nvidia cards is not a very high priority here. I think the opportunities in the Einstein code itself and in CUDA version selection probably very greatly outweigh any driver improvement opportunities.
I'm amused at the insistence on comparing single-instance processing time. If the interest is in Einstein productivity, that makes sense when considering users running pure defaults (which many do) but for the propeller-heads debating here, the sensible comparison is at best operating point.
Just because some of the AMD cards are not functional using current drivers and code at multiplicity greater than one is no reason to force the comparison to that point save for the purpose of falsely favoring AMD.
The 750 and 750Ti cards are remarkably good power efficiency for their generation. If the 1070 or 1080 can equal them that would be a considerable improvement over the Maxwell2 cards (not to mention any AMD card). We'll see. In comparing power it is important to select equivalent measurement. Even if power reporting from the card itself as displayed by a monitoring application were accurate (which in several cases I have observed not to be true), it does not actually incorporate the adders to power dissipation elsewhere in the user system associated with running Einstein work on the GPU, including CPU, motherboard RAM, motherboard I/O, and power supply imperfect efficiency losses. For that reason I think total box power measurements are greatly to be preferred.
Those 7970s may prove to be the best overall value still. You can get 150-160k daily from them at considerably less than their rated 300W TDP, and they or their 280x equivalent are easily obtained second-hand for $150.
This system pulls 80w idle at the wall with a 77w i5, two 140mm fans and a hyperactive Zalman cooler, 260w running 3 simultaneous E@H. It's slightly underclocked due to throwing errors at default clocks (the card was factory overclocked) so times are at 140k RAC equivalents:
This system pulls 80w idle at the wall with a 77w i5, two 140mm fans and a hyperactive Zalman cooler, 260w running 3 simultaneous E@H.
140k RAC equivalents:
That is impressive. I mentioned the power efficiency of the 750 cards previously. I have one dual-750Ti PC, which runs just under 140,000 credit/day at 173.1 wall watts operation, up from about 74 watts idle. These are overclocked beyond the EVGA "superclocked" point but have been stable for months at the current clock rates.
The cards in question (the EVGA "superclocked" variant) are currently available new for $110, or "used" from Amazon Warehouse Deals for about $97 each.
So compared to your 7970 reference, this dual 750 Ti system has somewhat higher purchase price, but considerably better power efficiency than the specific 7970 system you mention. On the other hand, two 750s uses up two slots, so with beefy power supplies and very generous case cooling one could stuff more compute power in a case using the 7970 option, which is an advantage in amortizing the system overhead over higher case level production.
Personally, I continue to think the 750 cards are great entry-level Einstein participant cards. Modest cost, small size means an easy case fit, surprisingly decent productivity (especially with some time invested in overclocking) and really almost wonderful power efficiency and very few would need to invest in a better power supply to use one. When Nvidia addresses that price point in the Pascal generation, they may greatly exceed the power efficiency, but probably not sooner. It is curious that Maxwell parts were rolled out bottom-up, while Pascal is coming out upper-middle first, then extending both down and up later, according to rumor.
This system pulls 80w idle at the wall with a 77w i5, two 140mm fans and a hyperactive Zalman cooler, 260w running 3 simultaneous E@H.
140k RAC equivalents:
That is impressive. I mentioned the power efficiency of the 750 cards previously. I have one dual-750Ti PC, which runs just under 140,000 credit/day at 173.1 wall watts operation, up from about 74 watts idle. These are overclocked beyond the EVGA "superclocked" point but have been stable for months at the current clock rates.
The cards in question (the EVGA "superclocked" variant) are currently available new for $110, or "used" from Amazon Warehouse Deals for about $97 each.
So compared to your 7970 reference, this dual 750 Ti system has somewhat higher purchase price, but considerably better power efficiency than the specific 7970 system you mention. On the other hand, two 750s uses up two slots, so with beefy power supplies and very generous case cooling one could stuff more compute power in a case using the 7970 option, which is an advantage in amortizing the system overhead over higher case level production.
Personally, I continue to think the 750 cards are great entry-level Einstein participant cards. Modest cost, small size means an easy case fit, surprisingly decent productivity (especially with some time invested in overclocking) and really almost wonderful power efficiency and very few would need to invest in a better power supply to use one. When Nvidia addresses that price point in the Pascal generation, they may greatly exceed the power efficiency, but probably not sooner. It is curious that Maxwell parts were rolled out bottom-up, while Pascal is coming out upper-middle first, then extending both down and up later, according to rumor.
What are your TIs clocked at? I have a pair of ASUS OC 750tis with dual fans and a 6-pin power connector that I've just bumped up another 100MHz each on GPU and memory for the first time, and they seem to be doing well. GPU is at 1315 (boost, 1171 baseline), memory 2800, 1.125v, 75% power.
They are fine cards, especially if you need to keep the heat down.
I too think the 750 Ti is currently the best performance/watt option.
My 750 Ti's have a 55k RAC and the GPUs consume only 30W.
Another advantage is that they run really cool so I don't need to switch them off during summer.
I just bumped the memory up another 100. Driver crashed about 5 minutes later. Oooops. I really should test this on a benchmarking program so I don't trash 10 hours of work.
Even though all four of my GTX 750Ti cards are the exact same EVGA SC model, no two arrived running the same voltage or default clock rate (!!) and none did I find myself clocking the same after my overclocking work.
As my overclocking tool has me expressing my request as a delta to default, I think it best to document the reported clock rates as given by GPU-Z.
In all cases I'm giving the long-term operating condition, not the highest I saw deliver a validated WU on a single trial. Of course, GPU-Z reports the memory clock a factor of two lower than do some other tools for the exact same operating point.
[pre]type # Core_MHz Mem_MHz
750 1 1416 1485
750Ti 1 1420 1645
750Ti 2 1413 1534
750Ti 3 1378 1504
750Ti 4 1346 1350[/pre]
Two cards on this list merit special mention: The 750 is a completely plain-Jane base model. Yet it overclocks with enthusiasm, and as overclocked is actually not much less productive than my 750Ti SC models. The fourth 750Ti was purchased quite recently from Amazon Warehouse Deals, and is by far the least capable overclocker of the four 750Ti cards of this model. In fact, I am actually currently running it at stock settings, as all my overclock tries failed, and my last tries were too little over stock to get me interested in looking for the last bit of juice.
Interestingly Ti card #4 came with the voltage (as reported by GPU-Z) set the highest, and thus was the only one to run into power limitation during my overclocking attempts. The behavior under power limitation was graceful--it just lowered the clock rate slightly dynamically.
While my overclocking gain was not so great on these cards as it was on the 970, it was much easier to do (no need to scratch one's navel while patting one's head in Nvidia Inspector) and the gain was in aggregate enough to notice.
RE: This night I was
)
My old 7970 GHz edition do single BRP6s in 2700s, no overclocking. I can only hope NVidia optimizes the drivers in a hurry.
RE: I can only hope NVidia
)
Einstein got a huge increase in productivity on all the Nvidia cards whose owners opted to allow Beta test applications simply by using the CUDA55 code generation path (as opposed to CUDA32). That has zero awareness of even Maxwell opportunities, and of course is only provided to users aware and interested enough to select the non default option. Clearly getting more out of the Nvidia cards is not a very high priority here. I think the opportunities in the Einstein code itself and in CUDA version selection probably very greatly outweigh any driver improvement opportunities.
I'm amused at the insistence on comparing single-instance processing time. If the interest is in Einstein productivity, that makes sense when considering users running pure defaults (which many do) but for the propeller-heads debating here, the sensible comparison is at best operating point.
Just because some of the AMD cards are not functional using current drivers and code at multiplicity greater than one is no reason to force the comparison to that point save for the purpose of falsely favoring AMD.
The 750 and 750Ti cards are remarkably good power efficiency for their generation. If the 1070 or 1080 can equal them that would be a considerable improvement over the Maxwell2 cards (not to mention any AMD card). We'll see. In comparing power it is important to select equivalent measurement. Even if power reporting from the card itself as displayed by a monitoring application were accurate (which in several cases I have observed not to be true), it does not actually incorporate the adders to power dissipation elsewhere in the user system associated with running Einstein work on the GPU, including CPU, motherboard RAM, motherboard I/O, and power supply imperfect efficiency losses. For that reason I think total box power measurements are greatly to be preferred.
Those 7970s may prove to be
)
Those 7970s may prove to be the best overall value still. You can get 150-160k daily from them at considerably less than their rated 300W TDP, and they or their 280x equivalent are easily obtained second-hand for $150.
This system pulls 80w idle at the wall with a 77w i5, two 140mm fans and a hyperactive Zalman cooler, 260w running 3 simultaneous E@H. It's slightly underclocked due to throwing errors at default clocks (the card was factory overclocked) so times are at 140k RAC equivalents:
https://einsteinathome.org/host/12254451
RE: Those 7970s This
)
That is impressive. I mentioned the power efficiency of the 750 cards previously. I have one dual-750Ti PC, which runs just under 140,000 credit/day at 173.1 wall watts operation, up from about 74 watts idle. These are overclocked beyond the EVGA "superclocked" point but have been stable for months at the current clock rates.
The cards in question (the EVGA "superclocked" variant) are currently available new for $110, or "used" from Amazon Warehouse Deals for about $97 each.
So compared to your 7970 reference, this dual 750 Ti system has somewhat higher purchase price, but considerably better power efficiency than the specific 7970 system you mention. On the other hand, two 750s uses up two slots, so with beefy power supplies and very generous case cooling one could stuff more compute power in a case using the 7970 option, which is an advantage in amortizing the system overhead over higher case level production.
Personally, I continue to think the 750 cards are great entry-level Einstein participant cards. Modest cost, small size means an easy case fit, surprisingly decent productivity (especially with some time invested in overclocking) and really almost wonderful power efficiency and very few would need to invest in a better power supply to use one. When Nvidia addresses that price point in the Pascal generation, they may greatly exceed the power efficiency, but probably not sooner. It is curious that Maxwell parts were rolled out bottom-up, while Pascal is coming out upper-middle first, then extending both down and up later, according to rumor.
RE: Meanwhile, Jeroen
)
See this thread on Nvidia memory clocks
RE: RE: Those 7970s
)
What are your TIs clocked at? I have a pair of ASUS OC 750tis with dual fans and a 6-pin power connector that I've just bumped up another 100MHz each on GPU and memory for the first time, and they seem to be doing well. GPU is at 1315 (boost, 1171 baseline), memory 2800, 1.125v, 75% power.
They are fine cards, especially if you need to keep the heat down.
I too think the 750 Ti is
)
I too think the 750 Ti is currently the best performance/watt option.
My 750 Ti's have a 55k RAC and the GPUs consume only 30W.
Another advantage is that they run really cool so I don't need to switch them off during summer.
-----
I just bumped the memory up
)
I just bumped the memory up another 100. Driver crashed about 5 minutes later. Oooops. I really should test this on a benchmarking program so I don't trash 10 hours of work.
RE: What are your TIs
)
Even though all four of my GTX 750Ti cards are the exact same EVGA SC model, no two arrived running the same voltage or default clock rate (!!) and none did I find myself clocking the same after my overclocking work.
As my overclocking tool has me expressing my request as a delta to default, I think it best to document the reported clock rates as given by GPU-Z.
In all cases I'm giving the long-term operating condition, not the highest I saw deliver a validated WU on a single trial. Of course, GPU-Z reports the memory clock a factor of two lower than do some other tools for the exact same operating point.
[pre]type # Core_MHz Mem_MHz
750 1 1416 1485
750Ti 1 1420 1645
750Ti 2 1413 1534
750Ti 3 1378 1504
750Ti 4 1346 1350[/pre]
Two cards on this list merit special mention: The 750 is a completely plain-Jane base model. Yet it overclocks with enthusiasm, and as overclocked is actually not much less productive than my 750Ti SC models. The fourth 750Ti was purchased quite recently from Amazon Warehouse Deals, and is by far the least capable overclocker of the four 750Ti cards of this model. In fact, I am actually currently running it at stock settings, as all my overclock tries failed, and my last tries were too little over stock to get me interested in looking for the last bit of juice.
Interestingly Ti card #4 came with the voltage (as reported by GPU-Z) set the highest, and thus was the only one to run into power limitation during my overclocking attempts. The behavior under power limitation was graceful--it just lowered the clock rate slightly dynamically.
While my overclocking gain was not so great on these cards as it was on the 970, it was much easier to do (no need to scratch one's navel while patting one's head in Nvidia Inspector) and the gain was in aggregate enough to notice.
Filipe wrote:Wondering if the
)
Polaris 10 leaked spec
AMD are announcing in a few hours, so we may be spoilt for choice very soon.