Reputation
Top tag
Next privilege 200 Rep.
See reduced ads
Badges
2 10
Impact
~11k people reached

  • 0 posts edited
  • 1 helpful flag
  • 2 votes cast
Dec
23
comment Intel Smart Response vs SSD as primary drive
@Moab I'd wondered the same but I suspect they're quite different. The Seagate cache is vastly smaller with greater knowledge of drive inner workings and real-time access to board hardware. Whether it makes much of a difference in practice or in addition to a second level controller is unclear to me. Conceptually, you could use it to unblock writes or cache the first few sectors of frequent accesses until the disk can catch up and take over.
Dec
23
comment Intel Smart Response vs SSD as primary drive
@AaronJSpetner I wondered the same but I don't know if anyone has the answer to this. I would guess not and unfortunately, this type of cache tends to do better with repeat accesses. The page file is a great candidate because you can be almost certain that every access is random and it is frequently accessed even when the machine has ample amounts of memory.
Dec
23
comment Intel Smart Response vs SSD as primary drive
@DavidSpillett It is different than using RAM, in that writes are persistent. That's significant because it means that writes don't need to block anymore. It's also considerably cheaper and therefore larger than RAM.
Jul
21
awarded  Notable Question
Apr
29
awarded  Notable Question
Nov
9
awarded  Popular Question
Aug
8
asked How big are copy on write pages in linux?
Apr
25
comment Is there a way to run a job exclusively on a chip?
Or if you want to target the average performance. Unfortunately, the application takes too long to run (hours to days) to directly measure it. Let's try to stick to the question asked. Q&A doesn't lend itself well to side discussions from speculation about the question.
Apr
25
comment Is there a way to run a job exclusively on a chip?
Edited to clarify
Apr
25
revised Is there a way to run a job exclusively on a chip?
added 170 characters in body
Apr
25
asked Is there a way to run a job exclusively on a chip?
Jan
29
awarded  Tumbleweed
Dec
23
awarded  Commentator
Dec
23
comment How can I measure speedups on my application when performance varies on my machine?
I don't think you understand the nature of the question. You use a profiler to identify a hot spot and come up a tweak to speed things up. A good software engineer will verify that the tweak does indeed speed things up and that's where I'm running into trouble. Running the same program multiple times can result in wide variations in performance, showing small slowdowns in some cases and big speedups in other cases so I can't test whether my tweaks are actually working! It's a problem I've noticed on our newer machines. Unfortunately, for some tweaks only the newer machines have enough memory.
Dec
21
comment How can I measure speedups on my application when performance varies on my machine?
Very little disk or swapping - it's a CPU bound application. No other processes running of significance. It occurs cache could be responsible but seems like a lot of variation for cache.
Dec
19
asked How can I measure speedups on my application when performance varies on my machine?
Nov
8
comment How can I fix this configure error?
Didn't look at config.log but have now. There was a suggestion buried in there that helped. Needed to point it to lib/pkgconfig. Never know how to read those log files and still don't but got enough out of it. Thanks!
Nov
7
asked How can I fix this configure error?
Mar
25
awarded  Popular Question
Jan
7
awarded  Supporter