anyone tried the SSE3 Linux version ?
wats the diff between static & dynamic ?
Printable View
anyone tried the SSE3 Linux version ?
wats the diff between static & dynamic ?
Fr3ak tried the SSE3 Linux version, said it needed work and benched lower than the last Linux.Quote:
Originally Posted by Hua Luo Han
Static doesn't change. Dynamic can change.
Ohh I understand now LOL. Thanks for replying.:DQuote:
Originally Posted by RyderOCZ
Sig system:
Benchmark results:
Number of CPUs: 1
2875 floating point MIPS (Whetstone) per CPU
11286 integer MIPS (Dhrystone) per CPU
Mates - FSC Amilo A64 3400+, 1Gb DDR333
Benchmark results:
Number of CPUs: 1
3686 floating point MIPS (Whetstone) per CPU
12045 integer MIPS (Dhrystone) per CPU
I pushed my FX a little harder and got
5416 fp/mps
17739 i/mps
@3.15 GHz
If it stays stable, I’ll go a little more.
EXCELLENT benches..is that on Linux?Quote:
Originally Posted by Thermodyne
LOL…….No it’s XP64. Right now I’m stuck with an old ASUS board, when I get my DFI fixed I should be able to turn it up some more with ease. This is one of the first FX57’s and it is a dog at the vcore the ASUS board gives it.
My 820 benches are also better on XP x64:Quote:
Originally Posted by Movieman
OLD (x86):
3438.89 million ops/sec
10797.97 million ops/sec
NEW (x64):
3781.82 million ops/sec
12711.17 million ops/sec
System in sig, both fresh installs.
If it was on Linux, his floating point would be north of 6000, with his integer around 18000 @ 3.15ghz.Quote:
Originally Posted by Movieman
Perhaps if it were you that was running it on NIXQuote:
Originally Posted by DragonOrta
If I was running it on NIX, it would be lucky to just be running.:eek:
All is well so far, so I pushed it up to 3.3GHz and got
5682 fp/mps
18647 i/mps :woot:
@3.30 GHz
That is definately one of the best Windows scores I've seen. :up: Sorry if it came across like I was trying to downplay your score, becase that was not my intention in the least.
Na, I wasn’t thinking that at all. I just don’t do much NIX. Supporting MS software pays my bills :)
Unfortunately only one of my living motherboards is semi-friendly to overclocking. And that one has crappy RAM & cooling! But I'm giving her all she's got...
Dual Opteron 252's @ 2743 Mhz Win Svr 2003 x64
4710 floating point MIPS (Whetstone) per CPU
15378 integer MIPS (Dhrystone) per CPU
X2 4400 89w @ 2409 Mhz Win Server 2003 x86
4131 Whetstone
13478 Dhrystone
X2 3800 @ 2196 Mhz Win XP Pro x86
3761 Whetstne
12321 Dhrystone
I predict we will overtake Free-DC by June 21st.:banana:
had a bit of downtime, but my Opteron is back up on a new mobo and performance is better too:
6/12/2006 1:58:08 AM|| Number of CPUs: 1
6/12/2006 1:58:08 AM|| 4795 floating point MIPS (Whetstone) per CPU
6/12/2006 1:58:08 AM|| 15645 integer MIPS (Dhrystone) per CPU
:)
my 2600+ mobile @ 2.38GHz does this.... its getting too warm to have it at 2.7GHz
http://i5.photobucket.com/albums/y18...6/Untitled.jpg
Sooo, that's what you got running.... I've got my eye on you. :flame:Quote:
Originally Posted by bobbywindel
Glad I can help. We need some good ol' fashioned competition around here. I might finally break down and get one of the DFI boards.
Benchmark results:
Number of CPUs: 4
4169 floating point MIPS (Whetstone) per CPU
13416 integer MIPS (Dhrystone) per CPU
Nothing earth shattering...;)
http://img461.imageshack.us/img461/5818/19532.th.jpg
5980 fp/mps
19532 i/mps
Here's something for the Windows people to shoot for.
Measured floating point speed 4859.12 million ops/sec
Measured integer speed 17371.43 million ops/sec
That's what I get for a 312*9 (2.81) 165
Benchmark results:
Number of CPUs: 1
3962 floating point MIPS (Whetstone) per CPU
12532 integer MIPS (Dhrystone) per CPU
Athlon64 3000+ @2500mhz
:slobber: What spped is that running at?Quote:
Originally Posted by Thermodyne
I was thinking at least 3300, thats a monster you have there!:woot:
That should hopefully do nicely for this weekend...
http://img68.imageshack.us/img68/7272/boinc8xt.jpg
Sweet Jeebus.....7K+ / 33K+ :eek: :eek: :eek: :eek: :eek: :eek: :eek:Quote:
Originally Posted by mikeguava
That will do do just fine :bows: :)
I just fell off my chair!!!! I think that will do Mike:) What speed is that at?
I really need new glasses, i'm seeing numbers that shouldn't exist.Quote:
Originally Posted by mikeguava
Now that`s one real benchmark for everyday use :D
Awesome, Mike :up:
Quick....have the defibrilator ready for when MM reads this thread.
We are on the phone now and I just told him:rofl:
Quote:
Originally Posted by Cooper
This actually is 24/7 setting - to make a crazy Benchmark I'll have to get the LN2 out....
Conroe on single stage....LN2 bench should be hopefully a bit more insane!
That was @ 4GHZ - loose memoryQuote:
Originally Posted by njkid32
He could also get strangled in his beard.Quote:
Originally Posted by RyderOCZ
Sweet Baby Jesus! 7K and 33K..I got to tell you with these guys I have to keep the smelling salts on the desk beside me!:DQuote:
Originally Posted by RyderOCZ
so thats 40K and times the 2 cpus is 80K+..Hmm..yea, I think 5000 points in a day is a fair mark for that machine!:p:
Now if it only had HT! :rofl:
Man you want everything:)
Opteron 165@2700
BH5 @2-2-2-5 250mhz
4658 floating point MIPS (Whetstone) per CPU
15218 integer MIPS (Dhrystone) per CPU
Guys, that is reality, every one of you will run similar speeds after 1-2 months :D
You score on 64bit linux would be way higher, this is what I got at only 3.1Ghz:Quote:
Originally Posted by mikeguava
http://www.xtremesystems.org/forums/...1&d=1149079115
Optimization made a world of difference in my benchmark scores
2882 Whetstone
5335 Dhrystone - Unoptimized scores for system in sig
Scores now are :
5147 Whetstone
16956 Dhrystone :D Much better indeed
intel t2400 stock
2691 whet
10256 dhry
anyone knows if that is about right
I am SOOOO glad Conroe is 64-bit. :D Now if Vapor put the E6700 he's got in 64-bit linux... :slobber:Quote:
Originally Posted by kiwi
Those are almost the exact same scores I get with a 144. 9x334 with 150 divider. BH-5 running at 250, 2-2-2-6. I wonder if your tccd was at 250 if your score would go down or not.Quote:
Originally Posted by LazyBum
4049/13201 prescott 2.4 @ 3.6 on air (see sig).
now I need that new psu...
My dothan 770 at 2567 gets 3000+/13,000+Quote:
Originally Posted by leviathan18
so I'd say thats a little low
welcome to the party Mat128..Glad to have you with us!Quote:
Originally Posted by mat128
my yonah is at 1.83ghz and i cant find how to oc it :( i think dothan and yonah are the same clock for clock
P4 530 HT 3.00GHz (stock - dell - no OCing :()
19/06/2006 08:51:33||Running CPU benchmarks
19/06/2006 08:52:32||Benchmark results:
19/06/2006 08:52:32|| Number of CPUs: 1
19/06/2006 08:52:32|| 3325 floating point MIPS (Whetstone) per CPU
19/06/2006 08:52:32|| 10850 integer MIPS (Dhrystone) per CPU
19/06/2006 08:52:32||Finished CPU benchmarks
thats not too bad I don't think. That is with the SSE2 patch.
Joshd, go into the bios and disable HT. Then go into windows and bench with the HT off you will see a diff:) When your done with the bench make sure you turn HT back on again before you run any WU's..
no can do... :( dell bios. you cannot configue much apart from the date, and boot order etc. no chipset options, no OCing, no temp reads... :( but yea, thanks anyway for the input.
i am building myself next time for sure. i have the dell becuase when i bought it i was not very good at computers, and would not have been able to build one myself, so it is fair enough i suppose...
would it be the same if i set the process affinity to one virtual core only?
EDIT: no it goes down a bit to:
19/06/2006 11:30:14||Running CPU benchmarks
19/06/2006 11:31:13||Benchmark results:
19/06/2006 11:31:13|| Number of CPUs: 1
19/06/2006 11:31:13|| 3304 floating point MIPS (Whetstone) per CPU
19/06/2006 11:31:13|| 10589 integer MIPS (Dhrystone) per CPU
19/06/2006 11:31:13||Finished CPU benchmarks
like that.
:O all this time i have been reading BOINC as BIONIC... hmmm...
Hey hey! It says number of CPU's 1... that means the HT is off!!?! They are not bad benches for a P4 3gig. they look like single thread benches to me!
Ewww I missed that good catch:)
hmmmm well the task manager gives me two CPU usage graphs, and i can set the affinity to CPU0 or CPU1... so i am assuming that it is on...
if it isnt, then i dont think there is any way i can actually turn it on.
Well I never bother to set affinity myself... so I'm not sure what way this is affecting the "Number of CPUs: 1"?!?. Like I have 3.4 northy and it'll say that I have 2 CPU's with HT on and 1 CPU with HT off! Maybe someome with more P4 rigs can jump in and sort this out! But like I said they are pretty OK benches for a 3.0gig.
EDIT: How many Rosetta instances are there in Task manager as a matter of interest??
only one. is there supposed to be two?
Well... now I ain't with my rig now, and am not familiar enough with the preferences pages for Rosey, but maybe you had selected just to use 1 CPU instead of 2!! (Windows and Rosey will see HT Proc as 2). Now if this is the case, I think NJKIDs advise on turning off HT for benches and turning on HT for running may have been carried out inadvertantly by your Rosey prefs.!! :cool: Now see if you can select from the preference pages to turn on 2 CPU's for rosey! You will get nice scores then!Quote:
Originally Posted by joshd
PS Anybody got better advice?? Now's the time to jump in!
got it, it was set to only use one in the preferences. thanks!
19/06/2006 12:00:57|rosetta@home|Scheduler request succeeded
19/06/2006 12:00:57|rosetta@home|General preferences have been updated
19/06/2006 12:00:57||General prefs: from rosetta@home (last modified 2006-06-19 12:00:36)
19/06/2006 12:00:57||General prefs: using your defaults
19/06/2006 12:00:57||Number of usable CPUs has changed. Running benchmarks.
19/06/2006 12:00:57||Suspending computation - running CPU benchmarks
19/06/2006 12:00:57|rosetta@home|Pausing task FRA_t327_CASP7_hom001_3_t327_3_1ub9A_IGNORE_THE_RE ST_162_724_2_0 (removed from memory)
19/06/2006 12:00:59||Running CPU benchmarks
19/06/2006 12:01:08||Running CPU benchmarks
19/06/2006 12:02:07||Benchmark results:
19/06/2006 12:02:07|| Number of CPUs: 2
19/06/2006 12:02:07|| 2610 floating point MIPS (Whetstone) per CPU
19/06/2006 12:02:07|| 5709 integer MIPS (Dhrystone) per CPU
19/06/2006 12:02:07||Finished CPU benchmarks
from what people have been saying, i take it that the bench goes down, but the actual crunching goes up?
still only one rosetta process though...
EDIT: two now, it DLed another task.
sorry for the crazy amount of posts i seem to be posting in the rosetta setion... lol.
Damn... I was afraid that it would rebench when you change Proc settings! i wonder is there away around that! Pity that damn dell doens't let you enable or disable HT! Is there any hacked off Bios that you can flash into that thing? Overall you'll get more points and more done with the HT on (and benched with it) but there is the opurtunity for the credits. ;)
no, no hacked BIOSs as far as i know.
is it better to have a higher bench is it? i mean, umm... not just knowing you have a good bench-wise?
Of course its better to have a higher bench... but you are asking me with a HT proc and the answer is a little longer. Your 2nd bench is lower then the first granted ... but it is not like 50%. Its 60%! Now instead of only one work unit been worked on at any given time, you now have 2 works units! Get my drift. Each one will get less credits BUT, comparitive data over time will always show that all things been equal, using HT on with associated benches will give you better scores.Quote:
Originally Posted by joshd
i see! so it will take longer to complete the WUs, but i am doing two at the same time, so in the long run, it is better.
have a banana for your helpful-ness. :D :banana:
Hmmm ... well.... you can set how long you want work units to work for. See .. the thing is you could, in reality... keep a workunit and do a VAST amount of work on it, and I reckon even beyond the constraints of BOINC. When you get a WU you are obliged to carry out the folding to get the first stage completed. Now for guys that like their own preferred lenght of time, you can set it to take 1 hour 2 hours.. etc etc in the preferences. However if the first stage of the folding takes longer than your preferences it will keep going until it is completed it first stage of folding. (I forget the correct terminology for 'stage'.) You can also I think set them going for 24 hours and that protein gets folded and folded and there are alot of predictions made on that WU.Quote:
Originally Posted by joshd
IN short, you can set the lenght of time to work on it, but remember your '1st stage obligations' have to be met. (UNless somebody changed the rules)
Laptop in Sig:
6/22/2006 6:47:45 PM||Benchmark results:
6/22/2006 6:47:45 PM|| Number of CPUs: 1
6/22/2006 6:47:45 PM|| 3070 floating point MIPS (Whetstone) per CPU
6/22/2006 6:47:45 PM|| 10799 integer MIPS (Dhrystone) per CPU
So is it better to have long WUs?Quote:
Originally Posted by riptide
Well, some people think its better to have them more than just 2 hours. You see, there is initialization period with every work unit.. lasts a couple of minutes. So if your WU's are short this initialization happens more than if your WU's are long. During the initialization you looze a little of time!Quote:
Originally Posted by mat128
Acer 8204 laptop w/ T2500 Yonah w/ SSE2 optimizations
(soon to put in my Merom 7400 - surprisingly the newest European Acer bios supports Meroms)
2944 floating
11394 integer
I just joined the XS Rosetta team and I should have a Conroe 6700 Stepping 5 and a Merom 7400 @ 2.6 up and running 24/7 within the next 1 week. It's great to see all the enthusiasm for these crunching causes!
:welcome: To the Team.Quote:
Originally Posted by Danger30Q
All i can say about your Conroes :cool: :woot: be sure to update us on the benchmarks when you get them.
3800+ at the new settings
Measured floating point speed 5449.36 million ops/sec
Measured integer speed 17873.63 million ops/sec
Intel Merom 7400 @ 2.8ghz (24/7 stable)
2 CPUs
Measured floating point speed = 4982.18 million ops/sec
Measured integer speed = 22994.21 million ops/sec
I have really struggled to get this chip stable at 2.9 which is what it's capable of but the board is holding it back. This box will be up 24/7 until the Conroes are released.
Is the Floating point score completly dependent on the CPU speed and nothing to do with what processor you have? It seems that a Opty 170, FX57, or X2 all running at the exact same speed have about the same Floating point score but have different Integer scores. Can someone tell me if I'm correct in this thinking? I haven't looked at too many Rosetta benches.
X2 4400+ @ 2.6GHz
CPUs: 2
4466 floating point MIPS
14583 integer MIPS
P4 550 @ stock 3.4GHz
CPUs: 2
3079 floating point MIPS
6284 integer MIPS
XP-M 2800+ @ stock 2.13GHz
CPUs: 1
2680 floating point MIPS
8791 integer MIPS
All more or less 24/7 Rosetta atm. Adding a Dothan soon and an E6700 later.
Thought I'd go ahead and update with my 64-bit linux Conroe benched at 3.38ghz.
Measured floating point speed 8193.98 million ops/sec
Measured integer speed 31016.07 million ops/sec
:eek: :eek: :eek: :eek: :eek: :eek: HOLLY HELL!!!! Man thats a great bench for that speed. Damn I need to goto linux!!!Quote:
Originally Posted by DragonOrta
http://img240.imageshack.us/img240/5...oebench4ke.jpg
Quote:
Originally Posted by njkid32
I told you njkid!
Dragonotra: Those are some crazy benches. Is it an e6600?
Yeah, it's a step 4 E6600, so this is about as far as I can go with my water right now. I use about 1.56v to get it benchable there. Once retail becomes available, I might try to sell this a get one of them, or just build another cruncher with a retail.
Ahem...so maybe this is a dumb question...but where are you guys getting Conroes???
Santa:) He lives in Virginia and Jim Carey was in his movie:rofl:Quote:
Originally Posted by mcflurry4321
For awhile before Intel stepped in and got upset there were lots of them in the For Sale section (you will need 100 posts to see). But now until the release of retail Conroe its gonna be a little hard to get them. So there were a few lucky ones that got some ES Conroes....
BTW :welcome: to Rosetta and thanks for helping out. You will notice that we kid alot here (me mostly) but when it comes to the important things you will see how serious we are as a team:) If you need anything please feel free to pm me or any of the guys here cuz everyone is more than willing to help out..
I'm not kidding when I say my conroe will be here July 12th ;)
I can't wait to see what this puppy will do!
Thanks a lot!!! I won't hesitate to ask any questions. I've been browsing the forums and everyone seems cool! Hope I can help out!Quote:
Originally Posted by njkid32
7/6/2006 7:05:38 PM||Running CPU benchmarks
7/6/2006 7:06:37 PM||Benchmark results:
7/6/2006 7:06:37 PM|| Number of CPUs: 1
7/6/2006 7:06:37 PM|| 4680 floating point MIPS (Whetstone) per CPU
7/6/2006 7:06:37 PM|| 15258 integer MIPS (Dhrystone) per CPU
7/6/2006 7:06:37 PM||Finished CPU benchmarks
On an A64 3400+ @ 2750
just noticed by NJ's and DO's benches,
their numbers are just about identical, slight edge in NJ's favor, but, basically even, and DO is 600MHz SLOWER
not bad @ all
DO I am going to mail you a raptor so you can install linux:rofl:
Yeah clock for clock there is no comparo with Linux (especially 64bit) to Windows. Just think if DO benched at the same speed I did. :eek:
Quote:
Originally Posted by njkid32
thats nice of you sending DO a new hard drive to run his rig on ;) :lol:
I was able to bench at 3.4ghz in Fedora Core 5, so I booted into XP and benched at 3.4ghz, just to see what a difference it really made.
Measured floating point speed 6127.14 million ops/sec
Measured integer speed 28317.65 million ops/sec
and the 64-bit 3.4ghz bench
Measured floating point speed 8285.44 million ops/sec
Measured integer speed 31175.42 million ops/sec
Funny you should say that. I've got an idle Raptor sitting on the lampstand next to me. The only problem is that it takes some work to get a linux install working on a computer with any hardware that's different from the original computer.Quote:
Originally Posted by njkid32
But Fedora Core 5 and Ubuntu are both pretty easy to install. And if you get in a bind, you might be able to give Ser a call. He has FC5 on some of his crunchers.
EDIT: Are you still in AZ? I keep thinking that you are in Vegas.
165 @ 2.9 Ram @ 242 2.5-3-2-5 ... Seems low to me doesnt it?
7/7/2006 2:07:09 AM||Benchmark results:
7/7/2006 2:07:09 AM|| Number of CPUs: 2
7/7/2006 2:07:09 AM|| 4881 floating point MIPS (Whetstone) per CPU
7/7/2006 2:07:09 AM|| 15977 integer MIPS (Dhrystone) per CPU
7/7/2006 2:07:09 AM||Finished CPU benchmarks
I found some driver issues in my Windows install and fixed them tonight... Here are my new benchies at 3987MHz 1.55v mem using 667 divider so it was in the high 500's at 4-4-4-8 2.2v... Still on air:)
http://img206.imageshack.us/img206/5...ebench24tx.jpg
Exactly what air cooling is it?Quote:
Originally Posted by njkid32
BTW, shouldn't you go to bed, LOL, it's at least 2am in the USA :p:
I am using a TT Big Typhoon with the stock fan on it. I will put a better fan on soon but I am going to run it on phase tomorrow I think:)
Only a Big Typhoon? OMG, those conroes will do fine on my Zalman 9500 :slobber:Quote:
Originally Posted by njkid32
Conroe 6700 ES @ 3.75ghz
Just installed Fedora Core 5 Linux-64 and it makes quite the difference:
BEFORE Linux-64:
Measured floating point speed 6627.41 million ops/sec
Measured integer speed 30633.12 million ops/sec
WITH Linux-64
Measured floating point speed 9078.58 million ops/sec
Measured integer speed 33338.27 million ops/sec
http://boinc.bakerlab.org/rosetta/sh...?hostid=270601
Over 9000 floating point is insane!
HOLY CRAP!:eek: Hey, post this in the main Rosetta section..You have bragging rights with this pal!:toast:Quote:
Originally Posted by Danger30Q
And if you look at my post at the top of this page, you will now see what I was so excited about back then. Awesome Numbers there, Danger30Q. :up:
Ok, I have been up all night with this:mad: but the results were well worth it:)
At 3876MHz or so I got--
Measured floating point speed 9417.6 million ops/sec
Measured integer speed 33839.27 million ops/sec
Not to shabby:) But, I am still working on a better oc its a pita cuz I have to do it all through bios not in Linux:(
see if you can break 10,000 on floating point:toast:Quote:
Originally Posted by njkid32
Ooops wrong bench:)
Measured floating point speed 9423.08 million ops/sec
Measured integer speed 35511.16 million ops/sec
IDK if I can break it cuz I think heat is an issue right now.
Put that sucker under phase! Don't let heat be your wall.
I know I am just being lazy:)
Oh I need to thank Fr3ak for helping me out with Linux:woot:
What distro are you guys using?? I heard that gentoo is the fastest distro as you can customize it for your comp.
I'm using Fedora Core 5 for now, just because Gentoo can be a b|tch to install
I hate Fedora by now. When I need a Linux as opposed to FreeBSD I use Debian. I am actively checking out Ubuntu. Didn't have time to mess with Suse but want to.Quote:
Originally Posted by moddolicous
well, gentoo really is a *insertbadwordhere* when you install it the first time. You have to built everything from scratch, deciding what you want. So you do only have what you need, no support in the programs for nonsense things.
Who needs a program, which got the gnome-gui routines in it, when you use kde. Also, who needs programs compatible with any 64-bit-cpu, when it could use especially the routines for your cpus.
Thats what gentoo is about. Choises.
If you wanna try out gentoo, i would be really happy to help.
But, back to topic: these scores really rock. i mean really really. :)