yeah, old generation cards are ok as well. doesnt have to be brand new ones.

power consumption isn't a super big issue within reason.

currently i'm running
5x EVGA 760 2GB
2x EVGA 750ti FTW 2GB
2x 1050ti SSC 4GB

was thinking about trying to phase out the 760s for some better cards. 1060's seem to be the way to go.

---

Seti@Home classic workunits: 29,492 CPU time: 134,419 hours

Looking at Shaggie76's GPU chart I notice that the 1050Ti turns out to be an interesting option. Performance wise it falls between the 960 and the 970 and yet is the most energy efficient of all the cards featured whilst still being reasonably priced.

---

And bonus, most 1050tis (minus the FTW model from EVGA) don’t require external power. All from the PCIe slot.

---

Seti@Home classic workunits: 29,492 CPU time: 134,419 hours

And bonus, most 1050tis (minus the FTW model from EVGA) don’t require external power. All from the PCIe slot.

But they can cause power problems with some motherboards when using multiple cards and are no good with dual rail PSU's.

Cheers.

And bonus, most 1050tis (minus the FTW model from EVGA) don’t require external power. All from the PCIe slot.

But they can cause power problems with some motherboards when using multiple cards and are no good with dual rail PSU's.

Cheers.

Hey, Wiggo: what do you mean by "no good with dual rail PSUs"?

---

Hey, Wiggo: what do you mean by "no good with dual rail PSUs"?

There are PSU's out there that have two 12V rails, 1 rail powers the motherboard while the other powers GPU's, so I'm sure that you can see where that leads to when you're trying to run GPU's with no PCI-e power connections with those types of PSU's. ;-)

Cheers.

Those are lower tier or much older PSU designs. Thankfully most modern PSU's have opted for a single rail +12V design.

---

Seti@Home classic workunits:20,676 CPU time:74,226 hours

A proud member of the OFA (Old Farts Association)

Those are lower tier or much older PSU designs. Thankfully most modern PSU's have opted for a single rail +12V design.

exactly. any modern PSU worth its snuff is going to be a single rail.

---

Seti@Home classic workunits: 29,492 CPU time: 134,419 hours

About the crunching performance x power draw x cost i made some tests and with the help of some data sended by Keith was abble to build this table. Obviusly the Power Draw & Crunching Times vary from the instant WU who is crunched so numbers are from an average of 10 WU for the Crunching Time & and instant picture for the Power so YMMV.

The test was made BLC WU only and both Hosts runs Linux CUDA90 builds and has 4 GPU's

Host--GPU-Crunching Time-Power Draw
Keith-1060 -165 secs---------88W
Mine-1060--176 secs---------86W
Mine-1070--127 secs---------117W
Keith-1080--100 secs---------135W

Now you just need to add the initial cost to know the real winner for the question: What is the Best bang -for-buck GPU.

Someone could add some data from the 1050Ti and the 1080Ti performance please?

---

Can't supply CUDA90 timing for my 1080Ti's but for SoG tasks you can add the 1080Ti at 160 seconds at 172W.

Calculated for two hosts over ten tasks average. Not as energy efficient obviously.

---

Seti@Home classic workunits:20,676 CPU time:74,226 hours

A proud member of the OFA (Old Farts Association)

Not as energy efficient obviously.

turn off that flux capacitor you have in there and maybe it will get better efficiency!!

---

Not as energy efficient obviously.

turn off that flux capacitor you have in there and maybe it will get better efficiency!!

My ideia was to compare only Linux boxes, we all know CUDA90 are a lot faster than the SoG builds.

---

I realize that. I would have to do some reconfiguring. It was a challenge to fit the 1080Ti hybrid into any computer because of its length. It is over an inch longer than a standard card or even a hybrid. I had to use Zalster's trick of moving the front mounted radiator fans outside of the case on my Air 540 cases. That is not possible on the Air 740 case with which my first Linux host is housed.

The only Linux host that could fit the 1080Ti would be the newest in the Core X9 case. I could swap one of its 1080 Hybrids for the 1080Ti Hybrid so see where it falls out under CUDA90.

---

Seti@Home classic workunits:20,676 CPU time:74,226 hours

A proud member of the OFA (Old Farts Association)

turn off that flux capacitor you have in there and maybe it will get better efficiency!!

My ideia was to compare only Linux boxes, we all know CUDA90 are a lot faster than the SoG builds.

See how he totally failed to see my comment, hahaha....

---

turn off that flux capacitor you have in there and maybe it will get better efficiency!!

My ideia was to compare only Linux boxes, we all know CUDA90 are a lot faster than the SoG builds.

See how he totally failed to see my comment, hahaha....

Nah . . . I need more POWER! I needed to dump that 21 gigawatts stored on the flux capacitor. I moved the cards so now Juan has some more data for his table.

Just have to keep looking in my rear view mirror to see if anyone is sneaking up on my tailpipe with a banana in hand.

---

Seti@Home classic workunits:20,676 CPU time:74,226 hours

A proud member of the OFA (Old Farts Association)

More data arrived...

Host--GPU-Crunching Time-Power Draw
Keith-1060 -165 secs---------88W
Mine-1060--176 secs---------86W
Mine-1070--127 secs---------117W
Keith-1080--100 secs---------135W
Keith-1080Ti--72 secs---------208W

Now we need someone who made the test with the TitanV,,1050 & 1050Ti to pick the winner.

---

To keep Titan V average times comparable to your someone could calculate an average from my completed tasks. Those having around 39 seconds are Titan V blc tasks. The power draw is 133-140W during the task. Those around 68 seconds are 1080Ti (202-225W) and 90 second tasks are 1080 (141-154W).

Petri

---

To overcome Heisenbergs:
"You can't always get what you want / but if you try sometimes you just might find / you get what you need." -- Rolling Stones

Thanks Petri, we not need to be so acurate the ideia is just a path to follow so...

With the new data....

Host--GPU-Crunching Time-Power Draw
Keith-1060 -165 secs---------88W (CUDA9.0)
Juan--1060--176 secs---------86W (CUDA9.0)
Juan--1070--127 secs---------117W (CUDA9.0)
Keith-1080--100 secs---------135W (CUDA9.0)
Petri--1080---90 secs----------150W (CUDA9.1)
Keith-1080Ti--72 secs---------208W (CUDA9.0)
Petri--1080Ti--68secs---------220W (CUDA9.1)
Petri--TitanV---39 secs---------140W (CUDA9.1)

The CUDA9.1 is faster then CUDA9.0 and that makes the GPU push a little more power as expected.
Very impressive numbers from the Titan V.
Can wait to see what mainstreet versions of this new generation of GPU's could do.
We going to need a bigger flux capacitor. LOL

Now we need just 1050 & 1050Ti to pick the winner.

---

Just for clarity, the 1080Ti report yesterday was on just BLC tasks since that is all I had. This morning I have Arecibo tasks for an average of 64 seconds.

---

Seti@Home classic workunits:20,676 CPU time:74,226 hours

A proud member of the OFA (Old Farts Association)

I am working on getting you that 1050ti information. What I need to know is the query in smi for the current pwr usage?

---

SETI@home classic workunits 4,019
SETI@home classic CPU time 34,348 hours