What's New: New Features in Version 23.x of Prime95
NEW FEATURES IN VERSION 23.x of PRIME95.EXE
New Features in Version 23.8 of Prime95
A communication bug, Primenet Error 1, was fixed in the case where proxy software
inserted the text "NetAnts: MaxConnection=1" in the HTML response from the Primenet
server.
Fixed memory allocation bug in torture test when using 926MB of memory on
the 20K FFT size.
The program will now try both full and a relative URL in trying to contact
the PrimeNet server. Some rare setups only work with full URLs and some only work
with relative URLs. The small downside is there might be an increase in network
traffic when the server is down. If this is a concern you can use the UseFullURL
option described in undoc.txt.
New Features in Version 23.7 of Prime95
A bug was fixed where the torture test used more virtual memory than necessary,
sometimes resulting in an "Out of memory" error.
New features in Version 23.6 of Prime95
Several changes have been made to the torture test.
Machines faster than 1 GHz will use new test data that runs a lot more iterations.
This will reduce time spent in initialization code, hopefully increasing stress
on the CPU.
The torture test stops after 100 ILLEGAL SUMOUT errors.
There are now 3 flavors of torture testing. One that operates on small FFTs
only, theoretically maximizing FPU stress. One that operates in place on large
FFTs only, maximizing heat and power drain. One that runs a blend of tests using
lots of memory, possibly catching errors that might only occur at a specific
memory address.
New features in Version 23.5 of Prime95
Four changes have been made after GIMPS' first false positive report in 7
1/2 years of operation.
The program now returns the number of errors that occurred when the result
is prime. A non-zero value will make us more suspect of the reported prime.
The save files will not be deleted. The user can then email these to me
and we can rerun the last 30 minutes of the LL test. It is hard to imagine a
second false positive report in this mini LL run.
The shift counter is now checked every iteration. If this variable AND the
FFT data was corrupted, then and endless loop of LL iterations generating zero
is possible - resulting in a false prime report.
Every iteration the FFT data is checked to see if the data has been zeroed.
The is-this-a-prime check now makes sure the FFT data is not NaN. NaN stands
for not a number and means the data is corrupt. The previous version checked
for zero and my C compiler returns TRUE for the test NaN == 0.0!
I restored the old behavior sending relative URLs. Some users had trouble
with this feature introduced in 23.4. So, UseFullURL=0 is now the default.
Some prefetching improvements were made for the Athlon, Pentium 3, and Celeron
2 processors. You can expect speed improvements between 3% and 10% for most FFT
sizes. Warning: the new code is slower for Durons and Celeron's with small L2
caches for FFT sizes 1024K and above.
New features in Version 23.4 of Prime95
Further performance improvements in the SSE2 FFT code. I've tried to do more
work while data is in the L1 data cache.
The program should now work with a "redirected proxy server". This was done
by sending full URLs in Primenet requests. While I don't anticipate any problems,
the old relative URL functionality can be had by setting UseFullURL=0 in primenet.ini.
A bug in the assembly language add and subtract routines was fixed. The bug
could affect ECM on exponents just below MAXIMUM_EXPONENT_FOR_FFTLEN - FFTLEN/2.
New features in Version 23.3 of Prime95
Further performance improvements in the SSE2 code for FFTs larger than 640K.
You should see about a 4% improvement in LL tests on a P4. However, FFTs between
40K and 512K might be a tiny bit slower.
Stage 2 of P-1 factoring has been recoded for more speed. WARNING: when continuing
from a save file created in stage 2 by a previous prime95 version, this version
will restart stage 2 from the beginning. ALSO: the stage 2 P-1 save file format
has changed - in dual boot scenarios you must upgrade mprime and prime95 at the
same time.
New features in Version 23.2 of Prime95
Further performance improvements in the SSE2 code. You can expect about a
4% improvement.
New features in Version 23.1 of Prime95
Big SSE2 FFTs now take the L2 cache size into account. P4 Celeron (128KB L2
cache) is faster for FFTs between 512K and 2M. P4 Northwood (512KB L2 cache) is
faster for FFTs larger than 1M.
Benchmark no longer times 256K and 320K FFTs, but does time 2048K FFT.
Support for torture testing FFT sizes from 1280K to 4096K added.
A 900 MHz P-III is now required to get first time LL tests by default.
Slightly faster SSE2 FFTs for lengths of 5*2^N and 7*2^N (e.g. 640K, 896K).