Hey Helheim Welcome to SoB.
Thanks for joining us
Hey Helheim Welcome to SoB.
Thanks for joining us
Yep. Had noticed him having a look around these parts. Hope the stickies were easy enough to understand.Originally Posted by bullet2urbrain
Hey Helheim Welcome to SoB.
Thanks for joining us
Welcome
312 Xeon Threads + GTX 1080 + GTX 1070 + BFL Monarch 700
Welcome !!
Yeap the stickies were great. When you have it set to not interact with the desktop there is no way to see what the clients are doing right? Are there any settings that should be changed from the default?
Thanks.
Yep. Had noticed him having a look around these parts. Hope the stickies were easy enough to understand.
okay well nevermind, your not welcome here anymore.
after you are storming past me and my lonely CPU.
I revoke your welcoming thread
Enjoy Helheim and thanks for the Powah.
I still have a long ways to go to catch up.
my mistake Helheim,
Sorry I am the lone user of CollegiateMafia , who is directly ahead of you... for now, the 19T in my sig is from what I crunched for XSTM
Settings -
You could up the priority to Normal in the client. (Or higher through Taskmanager if you wated to eek out extra few points, depending of course what you have it one and what else you wish to do with said rig.)
There are not a lot of settings really, however i do find that turning off 'transmit intermediate blocks' gives me a few extra points. Also if you choose not to use SBQueue and are connecting with the main servers with your blocks put in the IP address instead of spb.pns.net (or whatever it is) (saves looking up the DNS ie time). That'll eek out a few more points for us. I have the IP somewhere.. alternatively just do a dns lookup to get it yourself
What rigs you got going on this?
Interact with desktop -
Yep. Make sure to finalise you settings first on a visible client then restart the service if thats the way you will run this. At the end of the day, there isn't a lot to see with the client.
I do however make a ritual of restarting my clients eveyday anyway at least once. I had noticed (and posted somwhere too) that I had found my DX on a go slow. Still at 100% but obviously the OC must have fouled something up. Anyways a restart sorted it out.
PS: Second DX coming on line this evening. Can't wait to get home.
Last edited by [XC] riptide; 09-19-2006 at 07:19 AM.
Right now 4 instances on one DX rig and 3 instances on 2 others.
I tested the "intermediate blocks" setting a bit (24 hour test each). Doesn't make much difference on a rig that has multiple instances running since the the other 3 threads (DX rig) take up the slack when one is transmitting.
DeXter::2xXeon2.4@3200, GF6800, PC-DL, 1gb Centon, IWT+panaflo air, Win7
Quad::Q6600(g0)@3400(1.30V), GTX275, 965P-DS3, 4gb HyperX, Asus SilentSquare (air), Win7_64U
NuQuad::Q9400@3300(1.24V), 8800GT, G31M-ES2L, 2gb Patriot, stock air (bigger Q6600 one), WinXP64
Crunch::E4300@2400, onboard, GF7100PVT-M3, 2gb Patriot CL5, stock air, Vista64U
Does HT make a difference?
Thats on multiple instances yes. try on one instance.
Also when taking up the 'slack' you automatically fall into 2 clients per DX mode for that split second. Not ideal as 3 clients on all the time. Anyways thats my opinion.
Helheim-
Yes HT can make a difference. I need you to be more specific on the DX's. have they got 512/1/2 MB Cache?
This client need 512KB onboard cache (per client). You'll get away with 3 instances across 2 Xeons with 512KB cache each.
If you have 4 instances on 2 x Xeons w/512 cache you may be dipping into main memory... not ideal. I suggested before using 3 instead of 4 on KaptainBlazzeds DX. he came back and reported an overall 5% increase with 3 clients. (not 4)
In terms of HT, it is said that pinning 2 clients to one real Processor and letting the 3rd 'float' over the 2 logicals is ideal and the most aggressive usage.
Last edited by [XC] riptide; 09-20-2006 at 12:20 AM.
Ya they only have 512 cache. So would a Gallatin with 1 Mb L3 cache work better?
Tricky... cos its L3. It would be reaonable to suggest that if you HAD to run more than 3 instances, the 1 Mb L3 cache would probably be better. What L2 is in the Gallatin?
512 ... Ok so what are the most economical cpus/systems to use for SOB then?
Xeons and P4's / Core
AMD 64s do roughly equivalent in clockspeed to P4's. A 3800 X2 would be no slouch either!
The most economical? C2D!
A wolf in wolves clothing.
Kentsfield?
So which would be better:
Intel Core 2 Duo E6600 2.4GHz OR 2 Xeon 3.0 GHZ Irwindale 2M Cache
Do the E6600s have HT? Thanks.
6600 is all the output CollegiateMafia has right now.
thats an e6600 @ 3.4 Tight RAM 1:1
i believe the Xeon would be better.
I dont believe the Core 2's have HT.. if they do its not called HT
E6600 FTW.
a DX 3.0 will but out about 8 mill cem's/sec and a conroe @ 3.1 will put out about 12mill cem's/sec.
get the Conroe.
312 Xeon Threads + GTX 1080 + GTX 1070 + BFL Monarch 700
So Dempsey has HT but Woodcrest does not right but Woodcrest is higher FSB? I wonder how two Dempsey cpus would do?
What he said!
Posting Permissions
Reply With Quote
Bookmarks