BENCHMARKS 20x0!!!

The new 20x0(Ti) should be tested by the AnandTech-team or by the Tom's Hardware Page or the highly valued Swedish or Canadian site with the latest Linux CUDA version and compared against GTX like 980, 1080, 1080Ti and The TITAN V. And the results should be published.

I'll supply a version for the tests when the CUDA10 for the 20x0 is released and I get a Gpu(20x0Ti)card for development purposes for free or a loan. I'm not going to buy one in the future.

I bought one Volta (3k€) and I'm more than happy to buy some (n=3) used ones from the net. (reason: Perf/W)

--
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

even more sad that you lost an X79 board. they are expensive these days.

i want to get a new board for my test bench.
x79 = expensive MB + cheap ram + relatively cheap CPU
x99 = cheap MB + expensive ram + moderately expensive CPU
x299 = expensive everything.

---

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

When a motherboard offers me an auxiliary PCIe slot power plug, I always use it and plug in a cable. Just to reduce the strain the slots have to endure with 4 X 75w = 300w going through them. The PCIe aux power connectors are usually either 4 pin Molex or 6 pin PCIe connectors and their location varies all over the place on the motherboards that feature them.

---

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

A proud member of the OFA (Old Farts Association)

Thanks for the update Petri. Wasn't aware that there would be a CUDA10 release in conjunction with the Turing cards. I always read the compute tests in Anandtech's gpu reviews.

---

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

A proud member of the OFA (Old Farts Association)

When a motherboard offers me an auxiliary PCIe slot power plug, I always use it and plug in a cable. Just to reduce the strain the slots have to endure with 4 X 75w = 300w going through them. The PCIe aux power connectors are usually either 4 pin Molex or 6 pin PCIe connectors and their location varies all over the place on the motherboards that feature them.

it's also pretty uncommon to see a board with those extra PCIe power connectors. i've only really seen it on some EVGA boards. or mining motherboards.

i know this practice is uncommon here, but one advantage to hooking GPU's remotely via a USB3.0 type riser, is that you have to power the card's PCIe slot externally, taking all the power strain OFF the motherboard and directly from your PSU. so you wont burn up the board

---

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

When a motherboard offers me an auxiliary PCIe slot power plug, I always use it and plug in a cable. Just to reduce the strain the slots have to endure with 4 X 75w = 300w going through them. The PCIe aux power connectors are usually either 4 pin Molex or 6 pin PCIe connectors and their location varies all over the place on the motherboards that feature them.

it's also pretty uncommon to see a board with those extra PCIe power connectors. i've only really seen it on some EVGA boards. or mining motherboards.

i know this practice is uncommon here, but one advantage to hooking GPU's remotely via a USB3.0 type riser, is that you have to power the card's PCIe slot externally, taking all the power strain OFF the motherboard and directly from your PSU. so you wont burn up the board

Asus WS-series of mobos have them too.

---

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 for the update Petri. Wasn't aware that there would be a CUDA10 release in conjunction with the Turing cards. I always read the compute tests in Anandtech's gpu reviews.

Happy to deliver some news.
Cuda 10 will be available when the cards are too.
I'm not going to get it any sooner even though I'd like to.

--
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

The good news is, my 1200W PSU is fine, it is in my i7-960 with a 1080/1070/1060 now.

The board seems fine other than the 8 pin ATX connector, board under it is good.
I have a connector I removed from a board some time ago for this situation, but so far my attempts to remove the burnt connector have failed. I can't get it hot enough to de-solder it. Maybe today, or weekend.

Tried to take a pict, but my cam refuses to get out of playback mode.
My SDD (planned to move to the 960) took a hit. I must have laid the case down on it's back on a screw driver or something and the 'cage' around the SATA cable shattered, so just bare pins sticking out now. I will see it I can get an image off of it later.

I found an ASUS P9X79-E WS on eBay for $300 CAD from Canada even, not bad. Hopefully will be here around Monday.

I'm surprised I melted the connector on the P9X79-LE since that system only has 2 GPU on-board (1080 FTW Hybrid, ASUS 1070 Dual OC) and 2 EVGA 1060's on powered risers.

I was occasionally tripping my 810W UPS with it lately (always overloaded) so was on a 900W for the last 2 days of life. That was peaking at 960W when the 1080 was finishing tasks, ~860W normally.

I first ran across it on my ASUS ROG Crosshair V Formula Z board with my FX processors. Then got one on the ASUS X99-E-10G WS workstation mobo. I also see it is used on the ASUS Zenith Extreme X399 board I will likely be buying.

---

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

A proud member of the OFA (Old Farts Association)

I have a connector I removed from a board some time ago for this situation, but so far my attempts to remove the burnt connector have failed. I can't get it hot enough to de-solder it. Maybe today, or weekend.

Maybe this is old news to you, but whenever I have to remove a multipin connector and to prevent burning of the fiberglass board substrate, I always remove or snip off any of the plastic housing encasing the pins. That exposes each individual pin so that only heat needs to be applied to a single pin at one time. A good solder sucker or proper desoldering station is best also as trying to use solder wick normally steals too much heat from the joint to effectively clear the via anyway. So get those precision dikes out and snip away first.

---

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

A proud member of the OFA (Old Farts Association)

A loose pin equals a high resistance connection, which increases the amperage flowing through it. Which increases the heat, which increases the resistance >> ad nauseam >> = failure.

---

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

A proud member of the OFA (Old Farts Association)

Noticed in the description of the new MSI MEG Creation X399 motherboard for the new 32C TR2 chips, with their high power demands, that the EATX12V sockets are going to use solid pins instead of the normal Molex stamped and rolled Molex pins. A solid pin is able to carry a lot more current without overheating. Good design philosophy.

---

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

A proud member of the OFA (Old Farts Association)

even more sad that you lost an X79 board. they are expensive these days.

i want to get a new board for my test bench.
x79 = expensive MB + cheap ram + relatively cheap CPU
x99 = cheap MB + expensive ram + moderately expensive CPU
x299 = expensive everything.

The generic x79 MB's seem to be running $138USD but you need to find a "fast" shipper, not someone who sends it over a 3+ week journey. No these are not OC friendly as far as I can tell. But mine has been reliable (so far).

Tom

---

A proud member of the OFA (Old Farts Association).

they arent real x79 chipsets though. and they have a very real performance hit when used vs a real x79 board. i have one, but i wouldnt use it for something like SETI.

---

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

Hi Guys,
Anyone out there who needs more power for their video cards could try using one of these EVGA Power Boost .
Works if you don't have any auxiliary power ports on your MB, just plug into an empty PCI-E slot.
I have several of them. Need only one per computer.

---

Bruce

Didn't know about this product. Seems like a simple solution. Benefit is that it provides +12V power to the same common bus as the rest of the slots with the shortest distance also leading to minimal voltage drop. The only caveat is that I could see a motherboard having a bifurcated +12V power plane with the upper slots being fed by one bus and the lower slots being fed by another separate +12V bus. The Power Boost then might only boost some of the slots that are shared with the PCIe slot the Power Boost is plugged into.

---

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

A proud member of the OFA (Old Farts Association)

they aren't real x79 chip-sets though. and they have a very real performance hit when used vs a real x79 board. i have one, but i wouldn't use it for something like SETI.

Ok, I will bite. My Gtx 1060's on my ASRock(?) MB seem to be running noticeably faster than my 1060 on my generic X79 MB.

Recommend a middle priced X79? MB that will run an E5-2670 CPU please.
I may be able to afford one by Thanksgiving. :)

Tom

---

A proud member of the OFA (Old Farts Association).

Sorry. I meant CPU performance is less.

A GPU job might not be affected as much.

---

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

Sorry. I meant CPU performance is less.

A GPU job might not be affected as much.

OK. So far it looks to be a horse race in terms of Gflops for AVX and SSE4.1 on comparable cpus. I haven't run the dual CPU box long enough to say it really has "centered" at such and such time for the time estimates.

If my single CPU box were a bit faster it might turn out that AVX is in fact a few percent faster than SSE4.1 at least on the e5-2670's.

So could you tell me what MB's you are using for your higher core count systems? They all seem to be LGA 2011 v1/v2 cpus.

Thank you,
Tom

---

A proud member of the OFA (Old Farts Association).

Sorry. I meant CPU performance is less.

A GPU job might not be affected as much.

OK. So far it looks to be a horse race in terms of Gflops for AVX and SSE4.1 on comparable cpus. I haven't run the dual CPU box long enough to say it really has "centered" at such and such time for the time estimates.

If my single CPU box were a bit faster it might turn out that AVX is in fact a few percent faster than SSE4.1 at least on the e5-2670's.

So could you tell me what MB's you are using for your higher core count systems? They all seem to be LGA 2011 v1/v2 cpus.

Thank you,
Tom

Here is a $145 lga 2011 using the c602 chip set: https://www.ebay.com/itm/SuperMicro-X9DRW-IF-LGA2011-socket-R-Server-MB-w-RSC-R2UW-4E8-HEATSINKS/232889045875?hash=item363944d373:g:O58AAOSwe35bcx5z Its a dual, does that count against it?

Tom

---

A proud member of the OFA (Old Farts Association).