| Index | Recent Threads | Unanswered Threads | Who's Active | Guidelines | Search |
 |
World Community Grid Forums
Category: Retired Forums
Forum: Member-to-Member Support [Read Only]
Thread: Hyperthreading at the GSSC |
| No member browsing this thread |
|
Thread Status: Active Total posts in this thread: 5
|
|
| Author |
|
|
digitaloz
Cruncher Joined: Sep 7, 2005 Post Count: 6 Status: Offline Project Badges: 
|
New member with Hyperthreading workstations. How do you get this software to open up and use both processing streams?
|
|||
|
||||
|
retsof
Former Community Advisor USA Joined: Jul 31, 2005 Post Count: 6824 Status: Offline Project Badges: 
|
Actually, it is running at full power. Due to an Intel trick with your two virtual CPUs, it only reports "50%" with hyperthreading turned on. There are many threads about hyperthreading here that you could read, so I will not link to any one in particular. Search in the upper right box for "hyperthreading". Depending on your OS and other applications running, hyperthreading can be either on or off for best results. You can't run two WCG streams on the same machine. A few have done something with "fast user switching", but that's a whole other animal.
----------------------------------------
SUPPORT ADVISOR
----------------------------------------Work+GPU i7 8700 12threads School i7 4770 8threads Default+GPU Ryzen 7 3700X 16threads Ryzen 7 3800X 16 threads Ryzen 9 3900X 24threads Home i7 3540M 4threads50% [Edit 2 times, last edit by retsof at Sep 8, 2005 5:34:19 PM] |
||
|
|
RT
Master Cruncher USA - Texas - DFW Joined: Dec 22, 2004 Post Count: 2636 Status: Offline Project Badges: 
|
Please click here for the answer.
---------------------------------------- |
||
|
|
digitaloz
Cruncher Joined: Sep 7, 2005 Post Count: 6 Status: Offline Project Badges:
|
Yo dudes, I said I was a new member, not new to computers. The answers you gave and the information in that web link are so misleading. I wouldn't even want to put my name to it. These CPU's are spending vast amount of time doing nothing while they wait for the data bus to catch up. The whole point of hyperthreading was to keep the pipeline full of instructions while the data bus handled long memory I/O's. To say that the CPU is running at 100 % when the CPU processing level in the OS is reporting 50% is like saying, "Ignore the man behind the curtain". I was running BOINC on these workstations last week, and it allowed me to run two sets of tasks at the same time. The output of the two tasks in hyperthreading was about 35% greater then turning off hypertrheading and running a single task, and 48% faster than running a single task with hyperthreading on. My cooling fan for the heat pipe on this CPU hasn't event spun up to 70% while running this current project. Trust me, if my CPU was running at 100%, I could tell from the heat pumping out the back of my cooling tower. Just about all new Intel workstations are coming out with HT. If this project wants to make use of that kind of power, it's going to have to take the blinders off and get into the game.
|
|||
|
||||
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
[Grin] You have a point. So even though you're a new face 'round here, you're no pilgrim.
But joshing aside, there are complications. Intel has long been putting a lot of effort into architectural design for streaming media. They have had some success, but the jury is still out. Using hyperthreading for real work usually means using the streaming instructions with some careful programming that is very aware of the memory timing delays associated with the northbridge. Done just right, it works like a charm, giving the sort of figures you have used. Otherwise, hyperthreading is just a nice little speedup that lets you interleave a lightweight integer instruction thread with another thread for perhaps a 15% speedup. In particular, a thread using the standard FPU instructions will not play well with another similar thread. Once you get away from video editing, planning a program to do something like this leaves me nonplussed. Of course, the scientists planning these research projects are a lot smarter than I am, so I will not categorically declare that none of them will ever design a program for a project that could take real advantage of hyperthreading. But I would prefer to see any effort in this area directed toward true SMP programming for multi-processor systems. Oh, by the way . . . [said guilelessly] did you catch the mention in a column last week of some Intel engineers showing off some experimental silicon with a Pentium III descended core integrated with an on-chip memory controller? Just for engineering evaluation, not a proposed product. Short pipeline and no northbridge and with some additional video silicon integrated with the controller. I think that hyperthreading is not going to be the wave of the future. Just a guess on my part, since Intel keeps its own counsel. So this is all just guessing. The future will come soon enough and then we will know whether or not hyperthreading has a future. [I really thought that magnetic bubble technology would eliminate head crashes by replacing hard disks.  ]mycrofth |
||
|
|
|