Actually, a friend of mine is currently in the Technion doing a Masters'
degree, with the thesis subject being what is the best L1 etc. cache
remove policy that is best suited for SMT. As far as I know, this is, as
of yet, and unanswered question.
Shachar
Muli Ben-Yehuda wrote:
On
I think it's because Xeon processors use hyperthreading, which is sort
of like a dual processor inside one processor (not really, but a bit
like it). that allows instructions from separate threads to be processed
at once by the same CPU in one cycle.
You can read more about it here:
On Sun, Nov 10, 2002 at 10:48:50AM +0200, Boris Gorelik wrote:
I have even opened the box to verify the number of the CPU's.
Does anyone know anything about this behaviour? How should I treat the load
fugures I get from top?
Which kernel are you running? newer kernels (and newer CPUs)
Congratulations,
You just bought Intel Hyperthreading processors. Don't expect any earth breaking
performance from this (maximum 20% gain and even this is very rare)..
Thanks,
Hetz
On Sun, 10 Nov 2002 10:48:50 +0200, Boris Gorelik wrote
this is a VERY strange problem. My boss have bought a
[mailto:hetz;witch.dyndns.org]
Sent: Sunday, November 10, 2002 1:04 PM
To: Boris Gorelik; Linux-IL mailing list
Subject: Re: I have 2 spare CPU's (maybe not)
Congratulations,
You just bought Intel Hyperthreading processors. Don't expect
any earth breaking performance from this (maximum 20
On Sun, 10 Nov 2002 14:07:19 +0200, Dvir Volk wrote
AFAIK, the more you run multithreaded apps, the more performance gain
you get, isn't it?
On which linux apps should one see more imporvement? I guess servers
like apache and mysql can gain a lot - relatively - from
hyperthreading, for
On Sun, Nov 10, 2002, Hetz Ben-Hamo wrote about RE: I have 2 spare CPU's (maybe not):
As much as I know, Linux doesn't really excells in multi-threading (anyone -
please correct me if I'm wrong, I'm not very familiar on that issue)..
You CAN however do some testing with the latest multi
Dvir Volk wrote:
AFAIK, the more you run multithreaded apps, the more performance gain
you get, isn't it?
On which linux apps should one see more imporvement? I guess servers
like apache and mysql can gain a lot - relatively - from hyperthreading,
for example.
Something non obious to note about
Now, AFAIK Linux does have some scheduler code to handle this right, I
just can't seem to remember if it's in 2.4.x or only in 2.5.x
I think that today's it's inside RedHat's kernel version, not in the standard
Linus-releases version (I'm talking about kernel 2.4.x - not about 2.5.x)
I'll ask
On Sun, Nov 10, 2002 at 06:26:28PM +0200, Hetz Ben Hamo wrote:
Now, AFAIK Linux does have some scheduler code to handle this right, I
just can't seem to remember if it's in 2.4.x or only in 2.5.x
I think that today's it's inside RedHat's kernel version, not in the standard
Linus-releases
Hmmm. Then if the scheduler is unaware of SMT, then even on a
single-processor box SMT may degrade performance due to memory cache
issues -- when two unrelated threads are executed in parallel, the
effective size of the L1 and L2 caches is halved. With today's processor
vs. DRAM speed difference,
On Sun, 10 Nov 2002 14:44:39 +0200
Nadav Har'El [EMAIL PROTECTED] wrote:
1. Improving scalability: letting you run 10,000 threads concurrently, and
starting and deleting 100,000 threads per second, and things like that,
which I wonder if anyone really needs.
One of the most common
On Sun, Nov 10, 2002 at 06:52:02PM +0200, Eran Tromer wrote:
Hmmm. Then if the scheduler is unaware of SMT, then even on a
single-processor box SMT may degrade performance due to memory cache
issues -- when two unrelated threads are executed in parallel, the
effective size of the L1 and L2
13 matches
Mail list logo