GPU to 72 status...

[Rubiksmath]

Unrelated but I couldn't help but notice that 6M in general has a very high amount in the stubborn column, so much so that it looks out of place. I guess major P-1/ECM work has yet to be done there?

[chalsall]

Originally Posted by Rubiksmath

Unrelated but I couldn't help but notice that 6M in general has a very high amount in the stubborn column, so much so that it looks out of place. I guess major P-1/ECM work has yet to be done there?

Sorry, everyone... I'm *seriously* busy with deliverables...

Something changed on Primenet such that my spiders are no longer detecting Factors Found. Although, weirdly, the P-1 and TF levels appear sane. Except, for some reason, 0M, which gives an error.

Please give me some time to finish what I *have* to do. And then I'll drill down on this.

Fourth wall: At least Colab hasn't broken anything (more) recently... 9^)

[mrk74]

It's NOT causing any issues but found it interesting this started happening yesterday

[Uncwilly]

Originally Posted by Uncwilly

As of today, I have done 58 more exponents doing various 78->79 (36), 79->80 (15), and 80->81 (7)
Still no factors above 78.

I just looked at my numbers again:
I had a factor from 'computer' GPU_72 on June 29, 2022. Since then in the range from 332000000 to 339999937, I have 518 results from that 'computer'. Those range from 76->77, to 80->81. There have been no factors found since then. Based on the 1/bit level rule, I should have ~6.5 factors found. At some point either the factors should start rolling in or we should believe there is a problem with the system. Even using the 1/100 rule for higher bit levels, I should have 5 factors.
True randomness is lumpy, but how many standard deviations make something "impossible"?

Chris, am I an outlier in those using the GPU_72 notebooks?

[Mark Rose]

Originally Posted by Uncwilly

I just looked at my numbers again:
I had a factor from 'computer' GPU_72 on June 29, 2022. Since then in the range from 332000000 to 339999937, I have 518 results from that 'computer'. Those range from 76->77, to 80->81. There have been no factors found since then. Based on the 1/bit level rule, I should have ~6.5 factors found. At some point either the factors should start rolling in or we should believe there is a problem with the system. Even using the 1/100 rule for higher bit levels, I should have 5 factors.
True randomness is lumpy, but how many standard deviations make something "impossible"?

Chris, am I an outlier in those using the GPU_72 notebooks?

Working half outside of GPU72, I have 10 factors found in the last 1000 attempts in that range (54 factors in 5821 attempts over the past 105 days), with about half being on exponents with P-1 and an LL/PRP test. P-1 factoring had no doubt eliminated many exponents in that range, so my 0.9 to 1% success rate is about what I would expect.

I can double check your work if you can give a list of exponents and ranges (just copy it from your results page; I can wrangle the text). I have 25,000 GHzd/d available. I would start with the lowest bit levels.

[cmpute]

I'm unable to receive the confirmation email for registration, do you have any idea on this please?

[chalsall]

Originally Posted by cmpute

I'm unable to receive the confirmation email for registration, do you have any idea on this please?

To put on the table...

GPU72 is a hobby of mine. I really enjoy running it; it has helped me test various different things over the years. And I think it actually did some good with targeting resources when the wavefronts were really close.

But... I am not currently retired. I actually have to work. On some rather interesting and complex stuff (wouldn't come out of retirement otherwise).

I am aware of the SMTP issue, as well as the changes on Primenet which my spiders don't yet understand.

I just don't have any cycles at the moment...

I am working hard to have some time to invest in things like fixing GPU72. And, of course, I will be waking up early to watch Starship

I study Musk. I think he is brillaint. But I didn't realize just how insane he /might/ be until recently. Twitter?

[cmpute]

Sorry I didn't go through previous posts and didn't know the issue exists. I'm not pushing you for fixing this or anything, just take your time for whatever matters most to you :)

Is it possible to leverage Colab to do TF without a GPU72 account though?

[chalsall]

Originally Posted by cmpute

Is it possible to leverage Colab to do TF without a GPU72 account though?

Sure. Study the prior art... 9^)

[storm5510]

Originally Posted by chalsall

...GPU72 is a hobby of mine. I really enjoy running it; it has helped me test various different things over the years. And I think it actually did some good with targeting resources when the wavefronts were really close.

I had completely forgotten that GPU72 was your baby, so to speak. I haven't tested anything from there in years. From what I have read, the "powers" are getting out of my range. Recently, I have ran some of 76's, 77's, and just a few 78's. Nothing beyond. My aging hardware, including 2080, put me behind the times. I won't be doing much of anything for a while to come, at least, nothing of real value.

[Mark Rose]

Originally Posted by storm5510

I had completely forgotten that GPU72 was your baby, so to speak. I haven't tested anything from there in years. From what I have read, the "powers" are getting out of my range. Recently, I have ran some of 76's, 77's, and just a few 78's. Nothing beyond. My aging hardware, including 2080, put me behind the times. I won't be doing much of anything for a while to come, at least, nothing of real value.

A 2080 is still very useful. It's not as efficient as newer hardware, but easily beats the 1070s I only powered down last week.