[QUOTE=lycorn;389446]1) DCs are lagging first time LLs by an ever increasing amount.
2) First time LLs are the only type of work that can reveal a new prime.
3) 32 GB is great for P-1 (more than one test simultaneously)

Sounds conflicting, isn´t it? :confused:

If I were you, taking into account 1) and 3) above,[B] I would go for a mix of DCs and P-1s, [/B]also considering that P-1s will also help advancing first time LL work, by finding sometimes new factors and in any case offloading work from 1st time testers (no more need to perform a P-1 test before starting the actual LL).
All in all, do as you see fit... all contributions are valuable![/QUOTE]

Indeed. I am just wondering about the balance, which ATM is even. This has produced some interesting moments, such as the day recently that I found 2 DCTF factors and 2 P-1 factors waiting to be turned in. Given the items you list such as memory, plus my perception that the CPU does better when the same task type runs on paired integer cores; it seems that I'm in about as good a position as the circumstances allow. If I were to make any change, it would probably be to switch two workers from DC to P-1.

One more day of DC TF Processing and we have enough to cover the churn until recycling starts...:rogue:
(Nov 14 + 40days = Dec 24, 16 days left, current reserve of 20,000 ... )

If we keep up the last week's DCTF pace, we will finish all the DCTF in 8 or 9 months.

[QUOTE=Mark Rose;389549]If we keep up the last week's DCTF pace, we will finish all the DCTF in 8 or 9 months.[/QUOTE]

There might be some value in that. My vote would be to dedicate any "extra" firepower (that is after LLTF needs are satisfied) to finishing all the DCTF. It might be easier for us to only have one battle to wage. Easier to manage.

[QUOTE=TheMawn;389558]There might be some value in that. My vote would be to dedicate any "extra" firepower (that is after LLTF needs are satisfied) to finishing all the DCTF. It might be easier for us to only have one battle to wage. Easier to manage.[/QUOTE]

I'm going to continue to Let GPU72 Decide

[QUOTE=Mark Rose;389549]If we keep up the last week's DCTF pace, we will finish all the DCTF in 8 or 9 months.[/QUOTE]

I agree....actually you agree. I believe I posted the same about a week ago.:razz:

GPU72 Website slowdown
 

Any reason why GPU72 is so sluggish in the last couple weeks.....before that it was very snappy.

Internal Server Error
 

[QUOTE=petrw1;389580]Any reason why GPU72 is so sluggish in the last couple weeks.....before that it was very snappy.[/QUOTE]
It certainly has been slow for a while,
but the server seems to have hit rock bottom.

[QUOTE]The server encountered an internal error or misconfiguration and was unable to complete your request.
Please contact the server administrator, [EMAIL="chalsall@ideas4lease.com"]chalsall@ideas4lease.com[/EMAIL] and inform them of the time the error occurred, and anything you might have done that may have caused the error.
More information about this error may be available in the server error log.
Apache/2.2.23 (CentOS) Server at [URL="http://www.gpu72.com"]www.gpu72.com[/URL] Port 80[/QUOTE]

Let us all hope that Chris recovers fully soon, and not just because we are inconvenienced. :smile:

[QUOTE=chalsall;388628]So everyone knows, I'm moving house over this weekend, so I may be out of contact for a day or so. But will keep an eye on things from time-to-time by way of my Nexus.[/QUOTE]

[QUOTE=kladner;389590]It certainly has been slow for a while,
but the server seems to have hit rock bottom.

Let us all hope that Chris recovers fully soon, and not just because we are inconvenienced. :smile:[/QUOTE]

Related?

Running out of work. Changing my request to PrimeNet for a while till GPU72 is fixed.

[QUOTE=Uncwilly;389593]Related?[/QUOTE]

I had forgotten he was moving. I hope he is well otherwise.

[QUOTE=kladner;389605]I had forgotten he was moving. I hope he is well otherwise.[/QUOTE]

Hey All.

I'm in the new house, and ***finally*** have connectivity again (the pull-string broke on the first attempt pulling the fibre through; they had to dig up a section of conduit... :mad: :sad:).

I don't know why GPU72 was having issues, but as a quick-and-dirty fix I've restarted the database; it should be working now.

I have a LOT of email to go through, and "real work" to try to catch up on, but will be checking in here regularily from now on.

P.S. Went to my doctor yesterday to see about trying some other drugs to help with the "Chicken-shot-ya". I'm now into my sixth week.... (:mad: :sad: :pain:)

Edit: Awww, hell!!! I no know why GPU72 was having issues: a full partition. Stand by; working it...

[QUOTE=chalsall;389607]
I don't know why GPU72 was having issues, but as a quick-and-dirty fix I've restarted the database; it should be working now.
.[/QUOTE]

No error but not responding....

[QUOTE=petrw1;389609]No error but not responding....[/QUOTE]

Yeah... It's going to be offline for a little while longer. The partition holding the MySQL database filled up while I was offline (naturally).

Will post again when we're really back.

[QUOTE=chalsall;389610]Will post again when we're really back.[/QUOTE]

We're back. Sorry about that. As usual, a SSAE....

Got new assignments. Thanks!!

Get Assignments, DCTF, worked fine for me, just now (20:00 UTC), but the Available page did not fully draw the fields, and they did not populate. P95 tried to send 2 P-1 results, and they reported credit, but then the following came up-
[CODE][Tue Dec 09 14:00:28 2014 - ver 28.5]
Updating computer information on the server
PnErrorResult value missing. Full response was:
<!DOCTYPE HTML PUBLIC "-//IETF//DTD HTML 2.0//EN">
<html><head>
<title>500 Internal Server Error</title>
</head><body>
<h1>Internal Server Error</h1>
<p>The server encountered an internal error or
misconfiguration and was unable to complete
your request.</p>
<p>Please contact the server administrator,
chalsall@ideas4lease.com and inform them of the time the error occurred,
and anything you might have done that may have
caused the error.</p>
<p>More information about this error may be available
in the server error log.</p>
<hr>
<address>Apache/2.2.23 (CentOS) Server at v5.mersenne.org Port 80</address>
</body></html>
[/CODE]

[QUOTE=kladner;389625]Get Assignments, DCTF, worked fine for me, just now (20:00 UTC), but the Available page did not fully draw the fields, and they did not populate.[/QUOTE]

Could you give it another try?

Temporal wierdness,,,,

[QUOTE=chalsall;389626]Could you give it another try?

Temporal wierdness,,,,[/QUOTE]

Quick and complete, now. Thanks, Chris.

Sorry to hear that you are still afflicted.

EDIT: Unfortunately, P95 is still failing to communicate correctly with the server.

[QUOTE=kladner;389629]
EDIT: Unfortunately, P95 is still failing to communicate correctly with the server.[/QUOTE]

My three computers running Prime95 are also getting the same message

[QUOTE=Chuck;389649]My three computers running Prime95 are also getting the same message[/QUOTE]

Aw, shoot! Thanks for pointing that out; another "crashed" table. Fixed.

Thanks, Chris. P95 called home successfully for me.

[QUOTE=kladner;389679]Thanks, Chris. P95 called home successfully for me.[/QUOTE]

Just to take a short break from "life"...

George and I and others have talked about how to bring GPU72's functionality back into Primenet.

It will likely take some time, but please understand I'm 100% behind this (just in case anyone was wondering). GPU72 was always assumed to be temporary.

We now return you to your regular programming.... :smile:

:surprised:

[QUOTE=chalsall;389696]bring GPU72's functionality back into Primenet.
[/QUOTE]

:max:

[QUOTE=flashjh;389704]:surprised:[/QUOTE]

Just to clarify... GPU72 isn't going away; the "pretty graphs" will stay and assignments will still be available through it. But instead of GPU72 researving hundreds of thousands of candidates, it will instead be a proxy for Primenet, much like how LL and DC assignments now work.

I'm just giving you a hard time :-)

I'm sure everyone here understands!

[QUOTE=chalsall;389740]Just to clarify... GPU72 isn't going away; the "pretty graphs" will stay and assignments will still be available through it. .[/QUOTE]

Does "pretty graphs" include what is in mersenne.info?

If no .... such is life.

If yes there is the small issues of 0-1,000,000 and a couple other small ranges missing for the last few months.
And not only are the details missing but the total are also wrong with this data excluded.

[QUOTE=petrw1;389757]Does "pretty graphs" include what is in mersenne.info? [/QUOTE]

Yes.

[QUOTE=petrw1;389757]If yes there is the small issues of 0-1,000,000 and a couple other small ranges missing for the last few months. And not only are the details missing but the total are also wrong with this data excluded.[/QUOTE]

Yeah... (sigh) I know... (sigh)

I might have cycles to fix this next February. It's a "non-trivial problem space", caused by a very stupid programmer.... :wink:

[QUOTE=Chuck;389567]I'm going to continue to Let GPU72 Decide[/QUOTE]

So everyone knows, "LG72D" on the LLTF request form has been brought back into the LL range.

You guys hit this one out of the park! :cool:

I'm gonna unreserve the rest of my work since my second GTX670 seems to have gone flaky, won't start mfaktc unless I drop the clock speed from 1100MHz to 700MHz :sad:

[QUOTE=chalsall;389774]So everyone knows, "LG72D" on the LLTF request form has been brought back into the LL range.

You guys hit this one out of the park! :cool:[/QUOTE]

I might suggest two possible options for any excess Horse Power. ... there are many more.

1. Allocate it to DC-TF. As 3M (Mark, Mawn and Me) discussed in posts 3301 - 3304, DC-TF could be a thing of the past within a year leaving only 1 pool of effort to manage making it much simpler.

2. Become GPU to 75 in the LL Range

[QUOTE=petrw1;389779]I might suggest two possible options for any excess Horse Power. ... there are many more.

1. Allocate it to DC-TF. As 3M (Mark, Mawn and Me) discussed in posts 3301 - 3304, DC-TF could be a thing of the past within a year leaving only 1 pool of effort to manage making it much simpler.

2. Become GPU to 75 in the LL Range[/QUOTE]

Those of use who want to finish all the DCTF could leave our systems set to LG72D DCTF. I will for the time being.

[QUOTE=Mark Rose;389780]Those of use who want to finish all the DCTF could leave our systems set to LG72D DCTF. I will for the time being.[/QUOTE]

I will be switching my GPU processing back to protein crunching (Poem@ or Folding@) in Jan...) but until then I will leave them on LG72D LLTF .... but you can PM me if there is temporary shortage :-) (Hmm, Pretend I joined a chess team with LaurV )

[QUOTE=chalsall;389766]Yes.



Yeah... (sigh) I know... (sigh)

I might have cycles to fix this next February. It's a "non-trivial problem space", caused by a very stupid programmer.... :wink:[/QUOTE]
Great. That´s good news. I really like the info provided by GIMPS Visualization, and have been a bit disappointed by its lack of reliability over the last few months. Look forward to having it fixed.

[QUOTE=Mark Rose;389780]Those of use who want to finish all the DCTF could leave our systems set to LG72D DCTF. I will for the time being.[/QUOTE]

Makes perfect sense. :smile:

I will also remain on DCTF. Give us a shout if suddenly we had a need for LLTF.

I won't consider 75 bits to be a "need" until DCTF is finished.

[QUOTE=TheMawn;389805]I will also remain on DCTF. Give us a shout if suddenly we had a need for LLTF.

I won't consider 75 bits to be a "need" until DCTF is finished.[/QUOTE]

Me Three

M4. This is where your "5 years" go, Chris :razz:
I think you should step on your pride and make those cells white, now, hehe.

[QUOTE=LaurV;389833]M4. [/QUOTE]

Ok our Empire grows....can we total up how many GhzDays we have between us for our "Die! DCTF Die!!!" project .... anyone have a better name? I'm open.

AND ANYONE ELSE?????

I am committing 500 GhzDay's for the foreseeable future (again only as long as Chris doesn't plea for help at the LL front).

... and my sword ... and my bow ... and my axe!

[QUOTE=petrw1;389834]I am committing 500 GhzDay's for the foreseeable future<...>[/QUOTE]
[URL="http://www.gpu72.com/reports/worker/2423ae6e8f696d5e7d1447de91ca35a6/"]1T8[/URL] (this means 1800GHzD/D, and I meant it when I said it) as long as we have DCTF work and no requirement for other work, at least till Thai summer comes in March/April. :smile:

I'll be conservative and say that I will continue with ~460 GHz-d/d for the moment.

[OT]: I have been looking at PSUs, and ways of improving the PCIe power connections on the GTX 570, which is currently shelved. I think I either have to replace the connectors, or do something bizarre like cutting away the plastic on the existing connectors, and doing a solder buildup on the pins. The nominal cross section of those pins is square. On my 570, they are conical. I don't want to wreck another set of cable ends. However, the 570 could add another ~360-370, if I make it reliable again.
EDIT2: .....and get an adequate and efficient PSU to run an OC'd 580 and 570.[/OT]

I will continue with this DCTF as well for now - with averaging 300 GHzd/d.
The people saying their firepower in the last few posts would add up to 3060 GHzd/d. I calculated we are approximately 3088000 GHzd in front of our goal relating to [URL="http://en.gpu72.com/reports/current_level/"]this[/URL] table, or ~1000 days until we're finished.

[QUOTE=petrw1;389834]can we total up how many GhzDays we have between us for our "Die! DCTF Die!!!" project [/QUOTE]

~550 GHz from me.

Will any of the DCTF stuff also be brought to 72?

Assuming we're taking everything from 30M to 50M (inclusive) just up to 71 from where it is, we have roughly the equivalent of 420,000 69 to 70 jobs.

This puts my estimate somewhere along 1.3 million GHz-Days. If we have to do anything in the 50M range up to 72 bits then that will change significantly.


EDIT: My calculations took the GHz-Days from one of my ongoing 38M exponents as the base which might be a bit high. Taking into account the graph in manfred's post (I was looking for that one actually) which suggests that everything past 45M also goes to 72, then we're closer to 3 million GHz-Days, right about where his estimate is.


Add my GHz-Days and we're down to 2 years, 4 months.

[QUOTE=TheMawn;389871]Add my GHz-Days and we're down to 2 years, 4 months.[/QUOTE]

Take a look at [URL="https://www.gpu72.com/reports/estimated_completion/primenet/"]the second table on this report[/URL]. ~240 days to complete as the TF goal is currently configured (based on our thirty day rolling average of DCTF'ing).

However, as LaurV (correctly) pointed out, the "optimal TF level" for these ranges are actually higher for the higher ranges than shown, so we have some more work to do. I will adjust the report to be "correct" over this next weekend.

Also, keep in mind that you guys only seriously "surged" two weeks ago. The estimates given on the above report is based on the middle of the linear regression line on [URL="https://www.gpu72.com/graphs/dctf/month/"]this graph[/URL]. Read: If you guys keep this up, the estimate will continue to drop. :smile:

[QUOTE=manfred4;389842]I will continue with this DCTF as well for now - with averaging 300 GHzd/d.
The people saying their firepower in the last few posts would add up to 3060 GHzd/d. I calculated we are approximately 3088000 GHzd in front of our goal relating to [URL="http://en.gpu72.com/reports/current_level/"]this[/URL] table, or ~1000 days until we're finished.[/QUOTE]

Mine is usually 1200 GHz-d/d.

Speaking of rolling averages, would it be possible to have the GHz Days per Day on the /account/assignments page be changed to a rolling average instead of an account lifetime number? Perhaps along with the calculation used on the get assignments page, too?

Though I guess that's less relevant if the functionality is being moved to PrimeNet.

[QUOTE=petrw1;389834]…"Die! DCTF Die!!!"…[/QUOTE][YOUTUBE]trfg6WIDryQ[/YOUTUBE]

We may not have enough power to move all those ranges up as per those graphs. Additionally, it does not really worth to do so. The most work is done by top-of-the-line cards, and also we have to consider the P-1 done in the range, which decreases the chances of finding factors, from about 1 in 70 to about 1 in 90. All you guys doing DC here, check your history and see how many factors you found, for how much assignments (in bitlevels) taken, how long time it took, and how many LLDC tests could you do in this time.

No need to reply, it was just rhetoric.

As I repeatedly said (didn't you find out that I am playing Davieddy here, at least since he left? :wink:), the "optimum" level for the DC would still be according with[URL="http://www.gpu72.com/reports/current_level/"] this table[/URL] (i.e. 45M@72), except that the 56M has to be moved at 73 (i.e. the last 4 cells of the "72" column in the DC table, from 56 to 59 inclusive, have to be white, and not yellow). [B][U]That is all the change we need.[/U][/B] Of course, this changes Chris' internal kitchen too, it is not only about coloring the table, but about grabbing and releasing them too, at the right moment. But I am convinced he can deal with it, assuming he wants and has time.

We can get rid of DCTF once forever, if we stay on it next year. Actually, I would (optimistically) say half year... Maybe less...

[QUOTE=LaurV;389838][URL="http://www.gpu72.com/reports/worker/2423ae6e8f696d5e7d1447de91ca35a6/"]1T8[/URL] (this means 1800GHzD/D, and I meant it when I said it) as long as we have DCTF work and no requirement for other work, at least till Thai summer comes in March/April. :smile:[/QUOTE]

OK I will continue 1T3 DCTF Let GPU72 Decide until LLTF is needed.

[QUOTE=LaurV;389913]The "optimum" level for the DC would still be according with [URL="http://www.gpu72.com/reports/current_level/"]this table[/URL] (i.e. 45M@72), except that the 56M has to be moved at 73 (i.e. the last 4 cells of the "72" column in the DC table, from 56 to 59 inclusive, have to be white, and not yellow). [B][U]That is all the change we need.[/U][/B]

We can get rid of DCTF once forever, if we stay on it next year. Actually, I would (optimistically) say half year... Maybe less...[/QUOTE]

OK, this table has been updated. Taking 56M to 59M to 73 "bits" [URL="https://www.gpu72.com/reports/estimated_completion/primenet/"]only adds ~28 days of additional work[/URL]; ~258 days total (and dropping as the rolling average stablizes to the fire-power being applied). Note that 59M and above is already all at 73 or above.

[QUOTE=LaurV;389913]you guys doing DC here, check your history and see how many factors you found, for how much assignments (in bitlevels) taken, how long time it took, and how many LLDC tests could you do in this time.

No need to reply, it was just rhetoric.
.[/QUOTE]

My rhetoric reply: 22 / 1395 = .01577% = 1/63.4
Mostly 70-71; a couple dozen 69-70
Beating the odds.

[QUOTE=petrw1;389974]My rhetoric reply: 22 / 1395 = .01577% = 1/63.4
Mostly 70-71; a couple dozen 69-70
Beating the odds.[/QUOTE]

Hmm, since I don't know of an opencllucas, I could have done 31 dc in the last 26d, but found 134 of 13400 (92% @70-71)

Yeah, my last submission was 2/220.

[QUOTE=chalsall;389953]OK, this table has been updated. Taking 56M to 59M to 73 "bits" [URL="https://www.gpu72.com/reports/estimated_completion/primenet/"]only adds ~28 days of additional work[/URL]; ~258 days total (and dropping as the rolling average stablizes to the fire-power being applied).[/QUOTE]
Beautiful! Thanks a billion.
[QUOTE] Note that 59M and above is already all at 73 or above.[/QUOTE]
Guess who moved it? :razz:

[QUOTE=TheMawn;389986]Yeah, my last submission was 2/220.[/QUOTE]

Last 12 months - 995/61687 = 1.61%

Mixture of types from gpu72

62/3013 DC = 2.05% (many to 72, quite a few to 73 bits)

21/1741 LL = 1.2%

The other 57,000 were ranges picked by hand

[QUOTE=Mark Rose;388426]The defaults will be okay, but there's usually room for improvement.

The three that matter are:

GPUSieveProcessSize
GPUSieveSize
GPUSievePrimes

I would adjust them in that order. As soon as you see a GHz-d/day number you've run it long enough. CRTL-C and adjust. There are only a few values for the first two, and I'd adjust GPUSievePrimes by jumps of 5000 or so.[/QUOTE]

Thanks ...took a while to get around to it....but got me about another 3%
GPUSieveProcessSize: 16 to 24 to 32 (Max) got me about 1%
GPUSieveSize: 64 to 96 to 128 (Max) got me about 2% more
GPUSievePrimes: 82487 - 90677 got me about 0.2% ... any more made it slower.

Current thruput about 515 at about 18 degrees C.
Varies with temp; seems to self adjust; GPU temp is kept below 75 degrees C automatically.

[QUOTE=Gordon;390040]many to 72, quite a few to 73 bits
[/QUOTE]
Those are first time LLTF, with no P-1 done, for which a rate of 1 in 65 is normal. Same for the guys who replied before.
When I look to[URL="http://www.gpu72.com/reports/workers/dctf/"] this table[/URL], I see different things. (adding the 68-73 y rows, comparing to the number of factors found).
Anyhow, Chris already modified the table, for me the subject is closed (at least for a while :razz:).

hmm , is there a way to get daily stats Exact instead of adjusted?

[url]http://www.gpu72.com/reports/workers/dctf/day/[/url]
[url]http://www.gpu72.com/reports/workers/day/[/url]

[QUOTE=LaurV;390080]Those are first time LLTF, with no P-1 done, for which a rate of 1 in 65 is normal. Same for the guys who replied before.
When I look to[URL="http://www.gpu72.com/reports/workers/dctf/"] this table[/URL], I see different things. (adding the 68-73 y rows, comparing to the number of factors found).
Anyhow, Chris already modified the table, for me the subject is closed (at least for a while :razz:).[/QUOTE]

No, I did a LOT of DC work to 73 bits...

GeForce GTX TITAN Z
 

I am not into this subject so please excuse my question. How many GeForce GTX TITAN Z is needed for Trial Factoring be ahead of Prime95 LL wave?

[QUOTE=pinhodecarlos;390150]I am not into this subject so please excuse my question. How many GeForce GTX TITAN Z is needed for Trial Factoring be ahead of Prime95 LL wave?[/QUOTE]

Titan Z's are very inefficient on GHz/days per price, there are better options like the GTX 970 or R9 295... For the actual question, I'm not sure... someone like chalsall will know.

How many of these would help?

[URL]http://www.scan.co.uk/3xs/configurator/litecoin-mining-pc-10[/URL]

[QUOTE=pinhodecarlos;390163]How many of these would help?

[URL]http://www.scan.co.uk/3xs/configurator/litecoin-mining-pc-10[/URL][/QUOTE]

According to the figures from [url]http://www.mersenne.ca/mfaktc.php[/url], that rig could do 2,246.4 GHz-days/day.

[QUOTE=Mini-Geek;390172]According to the figures from [URL]http://www.mersenne.ca/mfaktc.php[/URL], that rig could do 2,246.4 GHz-days/day.[/QUOTE]

No harddrive though, and I'm not sure how 6 GPU's would work with mfakto... and the power consumption would probably be another consideration as well.

[QUOTE=kracker;390175]No harddrive though[/QUOTE]
It boots Linux off of USB, no hard drive needed.
[QUOTE=kracker;390175]I'm not sure how 6 GPU's would work with mfakto[/QUOTE]
It has options to choose the device when you run. I'd expect that you'd run 6 instances of mfakto, one for each GPU. Since the GPU can sieve its own candidates, there's no significant CPU load, so you could even do a little extra crunching on that.
[QUOTE=kracker;390175]the power consumption would probably be another consideration as well.[/QUOTE]
It comes with enough to power the GPUs under a full mining load: two 1000W PSUs. I would be surprised if mfakto had significantly different power requirements. Granted, it will put out a lot of heat and could draw enough power to trip some circuit breakers (if put on one circuit) and cost you a small fortune in electricity, but the rig itself is a solid unit suitable for crunching - just what it says on the tin.

[QUOTE=Mini-Geek;390177]It boots Linux off of USB, no hard drive needed.
[/QUOTE]

For a guy who has used only Windoze in his time, is Linux hard to figure out? I'm working on a box and I'm not sure if I want to go Windows 7 straight away or wait a bit until it mutates into the computer I want to actually use all the time. I suppose I would be running mprime and MISFIT (for GPU TF'ing)?

The mining rig looks pretty cool. It's quite an investment if all you plan on doing with it is crunching for GIMPS. Its work would certainly not go unnoticed. It all looks pretty good. If any amount of miners say it's good, it's good. Mining won't be different enough from GIMPS work one way or another that there would be any problems. Dual PSU's with that kind of power each is more than plenty.

As was already mentioned, the danger is the high power draw. Putting them on two circuits has the advantage of reducing the strain on that one circuit but you're running the risk of one circuit being shut off while the other is still live.

I've heard some horror stories about the "secondary" PSU (the one powering just GPU's) shutting down while the rest of the system (tries) to keep chugging along. Apparently the GPU tries to draw its power through the PCI-E slot instead (devices can actually do that; I'm not up to date on the specific but the HD 7750 and weaker GPU's could do it) and that more or less blew up a board. I'm sure some boards have current protection but maybe some don't?


All in all, it's a pretty safe thing that a lot of people have done. The unit is probably fine.

[QUOTE=Mini-Geek;390177]It boots Linux off of USB, no hard drive needed.
[/QUOTE]
Yeah, I saw that... and also I just noticed BAMT is part of it, so I guess you could run mfakto on it.

-Need to Read More

EDIT: Making the driver recognize more than 4 GPU's have been a hit and miss for AMD cards...

[QUOTE=TheMawn;390183]For a guy who has used only Windoze in his time, is Linux hard to figure out? I'm working on a box and I'm not sure if I want to go Windows 7 straight away or wait a bit until it mutates into the computer I want to actually use all the time. I suppose I would be running mprime and MISFIT (for GPU TF'ing)?[/QUOTE]

If you've never used Linux before, it might take a while to get used to it... but if you're only TF'ing with BAMT it's quite easy... assuming the drivers don't give you a hard time.
MISFIT is for windows only. You can use [URL="https://github.com/teknohog/primetools"]this[/URL] on Linux...

That system I linked can also boot from Windows 8.1 through a 60GB Corsair Force LS SSD.

[QUOTE=pinhodecarlos;390207]That system I linked can also boot from Windows 8.1 through a 60GB Corsair Force LS SSD.[/QUOTE]

Are you planning a nice Christmas gift, Carlos? :smile:

Luigi

[QUOTE=ET_;390211]Are you planning a nice Christmas gift, Carlos? :smile:

Luigi[/QUOTE]

Not yet Luigi but looking at different solutions....:tu:

I think GTX 970s would be much better: 30% less power consumption per card at stock clocks for not much more money.

[QUOTE=Mark Rose;390222]I think GTX 970s would be much better: 30% less power consumption per card at stock clocks for not much more money.[/QUOTE]

Hmm, you will have to keep in mind the next version of mfakto is 20% faster on the 280/290's by switch from 15 bit to 32 bit kernels...
So 290x @ 1000ghz vs 970, double the wattage, double the performance, only $100 more..

[QUOTE=NickOfTime;390223]Hmm, you will have to keep in mind the next version of mfakto is 20% faster on the 280/290's by switch from 15 bit to 32 bit kernels...
So 290x @ 1000ghz vs 970, double the wattage, double the performance, only $100 more..[/QUOTE]

That's quite an improvement! That brings the efficiency very close to a 970/980. How overclockable are the 280/290 for trial factoring purposes? I haven't paid much attention to AMD cards, with Nvidia working far better with Linux.

[QUOTE=Mark Rose;390224]That's quite an improvement! That brings the efficiency very close to a 970/980. How overclockable are the 280/290 for trial factoring purposes? I haven't paid much attention to AMD cards, with Nvidia working far better with Linux.[/QUOTE]

Well, I now have 2 models of 290x's
XFX DD with run a bit hotter but overclock a bit better, 1100-1150 @ 850ghz/d/d
and HIS IceQ X2 which run 6c cooler but I only clock to 1100 820ghz

I have all 4 running at 1080 777ghz/d/d

asus z87-ws i4770s 8GB antec HCP 1300w , quad 290x @ pcie 8x,
I tried the cheap pciex16 riser cables and they didn't work, but the 3M cables did...

Not sure what the performance diff would be between pcie 8x and a bitcoin pcie 1x system would be...
but I had a 3% drop in TF benchmarks when I switched from 16x to 8x...

[QUOTE=petrw1;389974]My rhetoric reply: 22 / 1395 = .01577% = 1/63.4
Mostly 70-71; a couple dozen 69-70
Beating the odds.[/QUOTE]

And a few days later; after about 1,700 results I continue to be at about 1/64

AND ...twice now I found 2 factors for the same Exponent. (I guess I have the option that says don't stop after the first factor is found).

[QUOTE=petrw1;390264]And a few days later; after about 1,700 results I continue to be at about 1/64.[/QUOTE]
Beginner's luck... :razz:
(kotgw!)

Most of GPU72 appears to be operating very sluggishly.

[url]http://www.gpu72.com/reports/available/[/url]
gpu72 reservations are shrinking
are you winding down the website ?
whats the best way to get DCTF work ?

[QUOTE=srow7;391041][url]http://www.gpu72.com/reports/available/[/url]
gpu72 reservations are shrinking
are you winding down the website ?
whats the best way to get DCTF work ?[/QUOTE]

There is a plan to merge GPU72's functionality into PrimeNet. It must be moving along well.

[QUOTE=srow7;391041][url]http://www.gpu72.com/reports/available/[/url] gpu72 reservations are shrinking are you winding down the website?[/QUOTE]

Not at all! As Mark stated, George, James, Madpoo and I are simply migrating GPU72 and Primenet such that Primenet will take on most of the backend responsibilities for assigning GPU TF'ing. GPU72 will continue being able to provide assignments (acting as a Proxy to Primenet), provide the "Pretty Graphs", etc. In fact, to the users there should be no apparant change.

[QUOTE=srow7;391041]whats the best way to get DCTF work?[/QUOTE]

Same as before.

[QUOTE=chalsall;391068]As Mark stated, George, James, Madpoo and I are simply migrating GPU72 and Primenet such that Primenet will take on most of the backend responsibilities for assigning GPU TF'ing.[/QUOTE]

OK, George and I are close to the point where we can start testing having GPU72 proxy to Primenet for TF assignments.

I'm wondering if I could ask for a couple of volunteers to help test this. Perferably at least one person who regularily gets manual assignments using the web form, and at least another using MISFIT and is willing to monitor things relatively closely to catch any SPEs.

Oh, and another thing... With the massive surge recently undertaken in DCTF, the fact that the Cat 4 "churner's" assignments have started recycling, and the fact we're going to try going to 75 in the LLTF range above (about) 66M or so, What Makes Sense should really be in the LLTF range.

Of course, everyone is (and will continue to be) free to do what ever kind of work they want.

[QUOTE=chalsall;391072]I'm wondering if I could ask for a couple of volunteers to help test this. Perferably at least one person who regularily gets manual assignments using the web form, and at least another using MISFIT and is willing to monitor things relatively closely to catch any SPEs.
[/QUOTE]

Works beautifully for me. Grabbed 5 default DCTF assignments and the preview showed me exactly what I was going to get.

I grabbed 20 "Makes Sense" DCTF but only up to 71 instead and it worked perfectly again, including the pesky preview pane.

[QUOTE=TheMawn;391085]Works beautifully for me. Grabbed 5 default DCTF assignments and the preview showed me exactly what I was going to get.

I grabbed 20 "Makes Sense" DCTF but only up to 71 instead and it worked perfectly again, including the pesky preview pane.[/QUOTE]

Thanks... However, the proxying isn't actually active yet; George and I still have a bit of tweaking to do -- possibly tomorrow.

Are you volunteering to be one of the beta testers?

New GPU72 assignment protocol...
 

[QUOTE=chalsall;391086]Are you volunteering to be one of the beta testers?[/QUOTE]

Sorry, I realize now that my statement was ambigious. I was asking for those willing to be beta testers of a new pathway. I should have been more explicit -- initially the proxying will be done for a limited number of users.

I sould (more accurately) ask that those willing to be testers please PM me your willingness, and with your Primenet Username if I don't already have it (what you use to log into Primenet with) -- this is needed for the proxying. VERY IMPORTANTLY, please don't send me your Primenet password -- I don't need it, and I don't want it! :smile:

For everyone else, please be aware that shortly "What Makes Sense" and "Let GPU72 Decide" will assign work which is logical to the project as a whole, regardless if the LLTF or the DCTF form is used. As in, if it is more logical to the project to do LLTF'ing work, the DCTF form will give LLTF assignments. WMS will continue to honor the "Pledge" level, LG72D will set it (probably to 75 for LLTF, 72 for DCTF).

The two different forms is actually a historical artifact which it's time to "refactor" out. DCTF was an after-thought, and my thinking at the time was that those who use the DCTF form only wanted to do DCTF'ing.

Those who do really only want to do DCTF'ing can choose any other option on the DCTF form, and you'll be given DCTF work. But, frankly, we're again so far ahead in DCTF that very little additional work is logical.

P.S. The new (anti-inflammatory) drugs appear to be helping! Up to 20 WPM! :smile: Unfortunately I only have another four days of them; we'll see how things go after that....

Is there any way to have a setting that:

1. Only hands out DCTF
2. At a GPU72/PrimeNet chosen bit depth?

I kind of want to "finish" all the DCTF.

[QUOTE=chalsall;391093]..........

P.S. The new (anti-inflammatory) drugs appear to be helping! Up to 20 WPM! :smile: Unfortunately I only have another four days of them; we'll see how things go after that....[/QUOTE]

I sure hope it does the trick. When you say, "another four days", is that due to limitations on how much of the stuff you can take, or some other cause? Is this anti-inflammatory a steroid?

Sorry for the probing questions. However, I am very curious about the disease, and your progress of recovery. I wish you the best possible outcomes. How is your girlfriend doing?

EDIT: I have about 6K GHz-days of DCTF in queue. When that is done, I'll probably go back to LLTF.

EDIT2: I notice that both GPU72 and PrimeNet are being unresponsive. I am presuming that might mean change is in progress!

What will be asked of the Beta Testers? I would like to continue focusing my efforts on DCTF unless there's an emergency of some sort (i.e. IBM devote all their spare cycles). If I can continue to do this but help in any way, I'd be glad. I just don't know what you need, yet.

GPU72 has been down for the last few hours.

[QUOTE=cardogab7341;391114]GPU72 has been down for the last few hours.[/QUOTE]

Sorry about that; back now. A weird memory-starvation issue....

[QUOTE=TheMawn;391110]What will be asked of the Beta Testers?[/QUOTE]

Not much; just keep an eye on things, and yell if things aren't working as expected.

[QUOTE=TheMawn;391110]I would like to continue focusing my efforts on DCTF unless there's an emergency of some sort (i.e. IBM devote all their spare cycles). If I can continue to do this but help in any way, I'd be glad. I just don't know what you need, yet.[/QUOTE]

Continuing on DCTF is fine; everyone's free to do whatever they enjoy.

Please be aware however that we are now again so far ahead in the DC range that any work you do won't be "used" for quite a number of months; the "churning" has stabilized. But again, the choice is yours.

Oh, and so you know, the new system will honour the WorkType setting (LLTF or DCTF) so long as you don't use the "What Makes Sense" or "Let GPU72 Decide" options.

[QUOTE=kladner;391106]I sure hope it does the trick. When you say, "another four days", is that due to limitations on how much of the stuff you can take, or some other cause? Is this anti-inflammatory a steroid?

Sorry for the probing questions. However, I am very curious about the disease, and your progress of recovery. I wish you the best possible outcomes. How is your girlfriend doing?[/QUOTE]

Limitations on length of time allowed. And yes, it is a steroid, which I assume is the reason.

Thanks for your concern -- this has been quite an experience! My girlfriend is slowly getting better as well.

[QUOTE=Mark Rose;391101]Is there any way to have a setting that:

1. Only hands out DCTF
2. At a GPU72/PrimeNet chosen bit depth?[/QUOTE]

Hmmm... Good point.

Perhaps I could set things up such that What Makes Sense honours the WorkType (DCTF in your case), but sets the Pledge level. As in, only "Let GPU72 Decide" chooses which range to do as needed.

Thoughts? The only significant change is WMS currently honours the pledge level.

[QUOTE=chalsall;391126]Thanks for your concern -- this has been quite an experience! My girlfriend is slowly getting better as well.[/QUOTE]

Glad to hear it!

[QUOTE=chalsall;391128]Hmmm... Good point.

Perhaps I could set things up such that What Makes Sense honours the WorkType (DCTF in your case), but sets the Pledge level. As in, only "Let GPU72 Decide" chooses which range to do as needed.

Thoughts? The only significant change is WMS currently honours the pledge level.[/QUOTE]

I like that idea. It may surprise a few people if you change it though. But once the integration with PrimeNet is complete with the current plan, there wouldn't otherwise be any difference between WMS and LG72D, would there?

78M 73-74 bits DC-TF just assigned....
 

Is that a "BUG" or a "FEATURE" :smile:

For now I changed MisFit to give me specifically DCTF - lowest Exponent and I got 41M 70-71 Bits even though there are still 40M 71-72 Bits available....and 41M 69 - 70 Bits.

Then I unreserved the 78M from GPU72 and deleted from the MisFitWorkToDo. It that the correct procedure?