Interesting, wonder how it would perform if he'd have waited a week or 2 and dropped in the 1080's instead? As it was a mfg sponsored rig, I can't imagine it would have been an issue, but maybe that is going to be part 2? ;-) Power? Oh, you _do_ have that small fission reactor in the basement, don't you?

---

Some new rumours.
How do a Pascal based Titan & GTX 1080Ti sound?

             Titan    1080Ti
CUDA cores   3,456     3,840

384 bit memory interface,
480-576GB/s bandwidth.


Tech report Pascal news

---

Grant
Darwin NT

Great catch that!

Liked the single precision performance for F@H

from: http://www.anandtech.com/bench/GPU16/1515


And while their full review isn't out yet I really liked their earlier statement of "marketing FLOPs being closer to real-world FLOPs this generation".

Strangely, no new drivers though (not really). So I got that one wrong.

A few more links (some may have already been posted):
NVIDIA Posts Full GeForce GTX 1070 Specifications
NVIDIA official specs for the GTX 1070
Whitepaper: NVIDIA GeForce GTX 1080

Also AMD Polaris mid-range cards getting announced on June 1st.

PS apologies for missing the 1080 launch but I've been DIYing, spring-cleaning and replanting my chile-peppers ;) Check Café for a bit more info.

Newegg is sold out of the 1080 even before they went on sale, lol...

Just a heads up that I'm noticing reviewers start to mention some thermal throttling with 1080 Founders edition under sustained loads at standard fan profiles. Upping the fan apparently handles that, at some amount of noise cost, as you'd expect.

Apparently Hybridizing the cooler works pretty well. Here's one I stumbled across where they applied the cooler off an evga hybrid 980ti: https://youtu.be/w-3fi1ovAP0 (sustains over 2100Mhz afterwards)

---

"Living by the wisdom of computer science doesn't sound so bad after all. And unlike most advice, it's backed up by proofs." -- Algorithms to live by: The computer science of human decisions.

Yes, we saw that.

Todderbert and I had a long discussion on why this individual was allowed to try and disassemble and reassemble this poor card.

It's obvious he's never done it before, wrong tools, wrong approach...

I cringed watching both videos.

It should be a video on how NOT to disassemble a graphic card.

But back to your statement on throttling.

No one has been able to duplicate the 2.1 GHz that was demonstrated in Austin. The current thought is it's a voltage issue, ie..not enough juice. So the question was raised, were they using a modified GPU with extra electricity in order to hit that 2.1GHz?

Guess we will find out soon enough...

Yeah those manky screwdrivers and ripping the heads off screws, lol
(at one point looked like he was about to damage some of the surface mount caps)

---

"Living by the wisdom of computer science doesn't sound so bad after all. And unlike most advice, it's backed up by proofs." -- Algorithms to live by: The computer science of human decisions.

Anndddd... They're off! Your search for EVGA GeForce GTX1080

Was wondering how long it would take (almost instantly) and how much more than retail (lowest upcharge as I write this is $250) for these things on Fleabay. Not surprised, at all, exact same thing as the Ticketmaster crapshoot when popular concerts happen, almost 0 chance of getting very good/great seats without having the privilege of paying a scalper who has his minions trolling the website as soon as the tickets are released. I'll choose to laugh a little as I pick up a (probably) 1070 when they are released in quantity in the coming weeks/months, for a much lower price point, with I presume higher performance. While it would be interesting/fun to have a new one to play with and get a feel for RIGHT NOW, it isn't worth a penny more than retail, and even at that I'd prob go for the non reference version. Been around too long and it isn't my first rodeo, so I'll be patient, unless something drops in my lap, of course. ;-)

---

Wow. Seems to mee that someone really hates eVGA, or nVidia, ... maybe because AMD suck ? Don't get it yet obviously.

---

"Living by the wisdom of computer science doesn't sound so bad after all. And unlike most advice, it's backed up by proofs." -- Algorithms to live by: The computer science of human decisions.

Wow. Seems to mee that someone really hates eVGA, or nVidia, ... maybe because AMD suck ? Don't get it yet obviously.

?

---

If you could get thru to EVGA's website, it's been up and down since 1 am California time.

Wow. Seems to mee that someone really hates eVGA, or nVidia, ... maybe because AMD suck ? Don't get it yet obviously.

?

Not you Al. Just my cynicism detector keeps going off, and I can't tell where the external sources versus the internal ones are coming from.

---

"Living by the wisdom of computer science doesn't sound so bad after all. And unlike most advice, it's backed up by proofs." -- Algorithms to live by: The computer science of human decisions.

If anyone obtains a 1080 (or even a 1070) and starts using it on this project, please let us know here. I'm interested (and probably many others!) to know whether the changes in the Pascal architecture mitigate the VLAR issue at all. If your computer(s) aren't hidden, we can have a look.

---

If anyone obtains a 1080 (or even a 1070) and starts using it on this project, please let us know here.

As to BOINC, I'm almost entirely an Einstein person now. I do have a 1080 on order (from B&H) but not much idea when it will arrive. I'd be surprised if it comes before late next week.

Should it happen that I get mine running before others here at SETI have generated reports, I'll give it a whirl here. Is the application of interest v8? And would I need to install optimized software for the results to be of interest, or is the stock application adequate to the purpose? V0.44?

From all reports so far, I suspect that the 1070 is likely to be a better cost/performance tradeoff than the 1080. If the 1080 is not catastrophic, I intend to get a 1070 and perform direct substitution in the same system, which could be an aid to comparison.

---

That would be excellent and thank you. Stock is fine so whatever the project installs automatically, as there's no known optimization specifically for the VLARs currently anyways.

---

Should it happen that I get mine running before others here at SETI have generated reports, I'll give it a whirl here. Is the application of interest v8? And would I need to install optimized software for the results to be of interest, or is the stock application adequate to the purpose? V0.44?

Hi Peter. Lunatics v0.44 will deliver one of the applications people will like you to test - that'll be the cuda50 option. But I also hope to have an Open Beta for Lunatics v0.45 running by next week, and that will also enable you to test the other one people are watching - OpenCL SoG r3430.

It would be great if you could use your meticulous reporting skills on both of those - first "as is", straight out of the box, and then again under the tuning guidance of Jason and Raistmer respectively as you all get the feel of the new architecture.

And here's a summary of the custom cards on their way.

---

Nice to see you again Archae! Thanx for dropping by :)

Any chance there's a DanNeely over @Einstein? Huge fan of his from the AnandTech forums (though I don't comment over there, just binge-read). I know he's a BOINCer... be nice to have his insight @SETI too :)

----

@Jason

When you say "throttling"... You mean from Turbo down to normal/reference clocks right? Or have u seen reports of clocks going down into the triple digits?

And here's a summary of the custom cards on their way.

Thanks for the link Dr Grey.

This one is interesting, 2-8pin.

As is this one here, 1-8pin.

Nothing seen from PNY or Zotac on the Hybrid front.
And yes I'd wait, places are charging double or even almost triple the $699 right now.

---

The T1 Trust, PRR T1 Class 4-4-4-4 #5550, 1 of America's First HST's

Should it happen that I get mine running before others here at SETI have generated reports, I'll give it a whirl here. Is the application of interest v8? And would I need to install optimized software for the results to be of interest, or is the stock application adequate to the purpose? V0.44?

Hi Peter. Lunatics v0.44 will deliver one of the applications people will like you to test - that'll be the cuda50 option. But I also hope to have an Open Beta for Lunatics v0.45 running by next week, and that will also enable you to test the other one people are watching - OpenCL SoG r3430.

It would be great if you could use your meticulous reporting skills on both of those - first "as is", straight out of the box, and then again under the tuning guidance of Jason and Raistmer respectively as you all get the feel of the new architecture.

Also very constructive musings going on in the background with Petri. Our somewhat more elaborate changes combined will probably take some more basic device parameters and performance characteristics to work out completely, So I'll have a small suite of unit tests accumulated before these devices reach heavy use.

---

"Living by the wisdom of computer science doesn't sound so bad after all. And unlike most advice, it's backed up by proofs." -- Algorithms to live by: The computer science of human decisions.