As long as you are using the same driver, this doesn't tell us much news.
Would anyone willing to test this with a newer driver that caused errors with BRP4?
BM
Yes, I can do that for HD 4600. Let me flush out a few BRP4 first, since we know they won't be happy.
Edit - I'll also do HD 4000, although that didn't suffer the same driver issues as 4600 with BRP4.
2nd edit - I see we have a new v10.18.14.4156 driver for the 4600, released 6 March. I'll test with that, unless anyone can suggest a reason for trying the older ones first? The 4000 is still at v10.18.10.4061
The wingman for your HD4600 has not completed his task yet, but we can already see your completion time dropped from 31 ks to 25 ks - very nice, if it validates!
The wingman for your HD4600 has not completed his task yet, but we can already see your completion time dropped from 31 ks to 25 ks - very nice, if it validates!
MrS
The HD4000, on the other hand, is possibly just marginally slower than before - not enough to worry about. Result due in about 30 minutes, but also with a slow wingmate. Follow-up tasks are running / ready to run, so we should get some validations - eventually.
I'm a bit nervous about that big speed-up on the HD4600 - have they optimised it by leaving something vital out? Well, only one way to find out.
Third HD4600 result is back - timing is consistent, but this is the first with a wingmate. And it's rated 'inconclusive' against a v1.39 NVIDIA GeForce GTX 670 with what looks like a very good record.
RE: As long as you are
)
Yes, I can do that for HD 4600. Let me flush out a few BRP4 first, since we know they won't be happy.
Edit - I'll also do HD 4000, although that didn't suffer the same driver issues as 4600 with BRP4.
2nd edit - I see we have a new v10.18.14.4156 driver for the 4600, released 6 March. I'll test with that, unless anyone can suggest a reason for trying the older ones first? The 4000 is still at v10.18.10.4061
Steady run of HD 4600 results
)
Steady run of HD 4600 results from the old driver at host 8864187
First task back with the new v10.18.14.4156 driver will be PM0009_010B1_222_3, in ~31,000 seconds - it seems to have started normally.
Equivalent links for HD 4000: host 5744895
First task with new v10.18.10.4061 driver PM0012_03631_326_1
Thanks for testing,
)
Thanks for testing, Richard!
The wingman for your HD4600 has not completed his task yet, but we can already see your completion time dropped from 31 ks to 25 ks - very nice, if it validates!
MrS
Scanning for our furry friends since Jan 2002
RE: Thanks for testing,
)
The HD4000, on the other hand, is possibly just marginally slower than before - not enough to worry about. Result due in about 30 minutes, but also with a slow wingmate. Follow-up tasks are running / ready to run, so we should get some validations - eventually.
I'm a bit nervous about that big speed-up on the HD4600 - have they optimised it by leaving something vital out? Well, only one way to find out.
There's also the issue that
)
There's also the issue that the valid/invalid ratio was about 1:2, so we need a few valid result to be sure enough (shall we aim for 5 sigma?).
MrS
Scanning for our furry friends since Jan 2002
RE: First task back with
)
The result files pass the initial check with the new, stricter validator, i.e. this result doesn't show the defects we saw with BRP4.
Looks promising so far!
BM
BM
PM0012_03631_326_1 is back
)
PM0012_03631_326_1 is back and valid.
Third HD4600 result is back -
)
Third HD4600 result is back - timing is consistent, but this is the first with a wingmate. And it's rated 'inconclusive' against a v1.39 NVIDIA GeForce GTX 670 with what looks like a very good record.
The HD4000 has thrown a
)
The HD4000 has thrown a validate error for PM0013_00241_382_1 - no sign of trouble in stderr_txt, to my eye.
Validate error on the
)
Validate error on the HD4600.
MrS
Scanning for our furry friends since Jan 2002