Encryption Projects as SU group

S

sdrfgs

Registered
Messages
139
Ok but keeping it simple

If you are only running 2 instances on the 3080 card you are not maxing it out at all, you can run a tool like GPUZ in the background and it will show this. ITs due to cudabiss not being very optimized.

consider this keep it simple for a test. On the 3080 run 4 instances at the same time.

perhaps 00-01 01-02 02-03 03-04 see what the final time is for this small range

Or try the bissbatch the other person uploaded, keep it simple one pc one gpu 4 instances for test purposes
 
I

ilovesattv

Registered
Messages
99
sdrfgs said:
can u post the input.txt for your 00 -ff test that you used
Click to expand...
So you want a copy of the input.text showing the text itself. that show 00000000 and FFFFF0000 from start to finish the search.
I use GPUZ and I think it says 85% used never 100% or even 90%
 
I

ilovesattv

Registered
Messages
99
sdrfgs said:
Ok but keeping it simple

If you are only running 2 instances on the 3080 card you are not maxing it out at all, you can run a tool like GPUZ in the background and it will show this. ITs due to cudabiss not being very optimized.

consider this keep it simple for a test. On the 3080 run 4 instances at the same time.

perhaps 00-01 01-02 02-03 03-04 see what the final time is for this small range

Or try the bissbatch the other person uploaded, keep it simple one pc one gpu 4 instances for test purposes
Click to expand...
Will do later when I have my time and not at 34 deg
4 Instances
1/ 00 to 01
2/ 01 to 02
3/ 02 to 03
4/ 03 to 04
 
D

dmr0x

Senior Member
Messages
315
moonbase said:
I assume that is USD ?
If there is a shortage and a Cuda Man can get one they could probably test it for a week or two then sell it on at the purchase price?

Scalpers were buying and selling at a profit a few years ago when the crytpto prices were high. Scalpers were waiting on d*****d for news of card drops at suppliers then getting family and friends to buy to sell on at a profit.
Click to expand...
yeah, expect the $ to be replaced with €/£ Its more expensive than the 4090 was at launch so best bang for buck would be 2 used 4090's
 
D

dmr0x

Senior Member
Messages
315
ilovesattv said:
Tried it and little gain. Could you images loading 16 instances on 4 computers. I go MAD with my damaged head.
Click to expand...
Nilaoda posted an automated solution to create the inputs to save alot of copy and pasting, check afew pages back.

​

 
M

moonbase

VIP
Donating Member
Messages
706
dmr0x said:
yeah, expect the $ to be replaced with €/£ Its more expensive than the 4090 was at launch so best bang for buck would be 2 used 4090's
Click to expand...

It has been that way for the last few generations of NVIDIA cards.
The CudaBISS processing speed per $ is better with two cards of the previous generation compared with one card of the new generation.

There are a few dreamers trying to offload RTX4090's in advance of the launch of the RTX5090's. They are asking ridiculously high prices in the hope of catching a toby and pick his pocket.
Once the RTX5090 is released on sale I think £850 to £950 will represent buyers value for an RTX4090. I spunked up £1175 collected for a 2nd user RTX4090 this time last year, one year on the price should be perhaps £1000 +/- With the launch of the RTX5090's this should dip below a bag of sand.
 
D

dmr0x

Senior Member
Messages
315
moonbase said:
It has been that way for the last few generations of NVIDIA cards.
The CudaBISS processing speed per $ is better with two cards of the previous generation compared with one card of the new generation.

There are a few dreamers trying to offload RTX4090's in advance of the launch of the RTX5090's. They are asking ridiculously high prices in the hope of catching a toby and pick his pocket.
Once the RTX5090 is released on sale I think £850 to £950 will represent buyers value for an RTX4090. I spunked up £1175 collected for a 2nd user RTX4090 this time last year, one year on the price should be perhaps £1000 +/- With the launch of the RTX5090's this should dip below a bag of sand.
Click to expand...
CEX have jacked there prices, some used costing more than New. 🤣
FPGA was way faster 10 years ago than any card today. GPU's are a sledgehammer to crack a nut approach. If you want to run Crysis though a FPGA is no use :)
 
S

sdrfgs

Registered
Messages
139
ilovesattv said:
So you want a copy of the input.text showing the text itself. that show 00000000 and FFFFF0000 from start to finish the search.
I use GPUZ and I think it says 85% used never 100% or even 90%
Click to expand...
Yes post the input file that you used and ill do some testing too
 
I

ilovesattv

Registered
Messages
99
dmr0x said:
Nilaoda posted an automated solution to create the inputs to save alot of copy and pasting, check afew pages back.

​

Click to expand...
I like to know this because me being a Stroke old man it does confuses me a lot does the same thing multiple times Normally if I use 3 Computers (to hell with 4 computers that 8 instances) If use 3 computers and have 2 instances on each and that's a total of 6 instances I would like to do it easier than what I do now
 
I

ilovesattv

Registered
Messages
99
I did a test on my last computer (the fast one) with the 3080 ti GPU and better motherboard and I found this. Normally on all computers I use 2 instances only from range 78-FF. Remember the other 2 computers have their own ranges and I start them at 09 to whatever. The reason why I start from 09 is that ive never had a key under this in 5 years. The closest was 11.

The photo below shows one instance on this fast computer, well fast enough for me. 8 hours 29 minutes for the full 78-FF in one instance. Now thats slow for that range so i decided to do 2 instances after this.


The photo below shows you the one i normally use which uses two instances which is at an average of 5 hours 56 minutes each. Remember if I use 4 computers which is a pain for me because I have to load 8 instances and my brain does not hack this, I can bring it down to 4 hours 50 odd minutes. Now this if 4 hours and 50 minutes and the other computers will be the same as they have been spaced so that the fastest computer does more and the slower computer does less thus they will all finish all around 4 hours 50 minutes give or take 2 minutes or so because of whatever stream I am using. Once again 4 computers, 8 instances is a headspin for me. I prefer 3 computers and 6 instances. On the 3 computers which are the fastest and have the best GPU's. Id like to find out how to simplify a solution to create the inputs to save alot of copy and pasting. But for me I do it manually, ugghhhhh.


The photo below shows four instances which I have never used. Now this is slightly faster at between 4 hours and 5.5 hours all because I took my time to start the four of them together. If I can select CudaBiss Watcher the four of them together, you will find them closer but if i let it go for half an hour the whole four will balance out. Same as the instances above with 2 instances. It is impossible to start the two together. The second one will always be slightly slower off the mark but once again after twenty minutes to half hour they catch up and both will be the same plus or minus a minute.


Take note, Look at the GPU-Z - GPU Load. You will notice that a single instance is 79%. Two instances is 87%. The four instances is 94% average.
 
Last edited:
S

sdrfgs

Registered
Messages
139
your cudabiss watcher is still using the old slower cudabiss

If you are manually starting each of the 4 instances then yes they will all have different timings. I would suggest making a batch file to start them all at the same time however much better to use the setup Nilaoda posted with the more optimized cudabiss newer release.
 
S

sdrfgs

Registered
Messages
139
@nilaoda
from the gui can we enter a locally selected input.txt for when we don't have the .ts recording? also some default key step sizes would be usefull to know. For various cards there could be some suggested defaults,
 
I

ilovesattv

Registered
Messages
99
sdrfgs said:
Is that all the options? perhaps you can post a few more details listing ALL the option commands? are you running 4 copys split over the full search range.
e.g input is 00 to ff ?
Click to expand...
For the 3 computers Using (4 computers is a pain in the arse) using 2 instances per computer total 6 instances to search the whole of my range from 09 to FF can search it it and get the key in the max of 5 hr and 50 minutes. Because all computers are different (Different GPU's and Speed of the Motherboard + Memory) I have made them to start and finished within, under 10 minutes and I do this because of a Table I have made. (picture coming soon.)
Now depend of my damaged stroke brain I can use the forth computer and can run my full range (09 to FF, Our way never seen a key under 09 so what start it at 00) and with 4 computers going that 8 instances I can search the the total range in 4 hrs and 50 odd minutes.
I have tried to simplified my system because using 8 instances ( 4 computers) or even 6 instance (3 computers) is a pain in the arse (My bottom, just in case you dont know what arse is). Have done this on paper format.
Maybe it good for a young smart man but not a old man that has had one Heart attack, Plus 3 full blown strokes and 8 TIA (transient ischemic attack, a small stroke) in 18 years. but I keep so far soldier on.
 
I

ilovesattv

Registered
Messages
99
This is what i had to do to work out all the different ranges that 3 or 4 computer than can give me. All I have to do is change one GPU, one Motherboard with different speed, one anything and the starts are the roughtly the same (can control this as I go a fast than I can do) but they finished differently. "eg Change the GPU from RTX3060ti to RTX 3070 that has a a few more cores and when it come to the end of its RANGE its finished early so it balance time so all the instances finish the same within 10 minutes

Don't Laugh but it works for me.
 
I

ilovesattv

Registered
Messages
99
sdrfgs said:
your cudabiss watcher is still using the old slower cudabiss

If you are manually starting each of the 4 instances then yes they will all have different timings. I would suggest making a batch file to start them all at the same time however much better to use the setup Nilaoda posted with the more optimized cudabiss newer release.
Click to expand...
Nilaoda work is good but I have to get my old head around this, Plenty of tea and scratch my head time.
 
You must log in or register to reply here.
Top