Tweet
I have read the treads on pegged CPU's. Is there a resolution to this issue. With me it is either 6.0.1.1 with a pegged CPU and correct volume or 5.0.1.4 with incorrect volume and no pegged CPU.
Thanks.
Thanks.
-
-
The eSignal datafeed brings more data to QC6 so the only solutions to a pegged CPU are to reduce the number of monitored Symbols and/or Studies until the CPU is no longer pegged. Running QC6 on a multiple CPU system will help a great deal but possibly not to the extent you need.
~Bob -
Thank you for the solution Bob. But this brings up another issue. If I have symbols of 500 max and I cannot view all of them due to my CPU being pegged then what is the point of using the E-signal feed?
Anyone is free to answer this question.
Thanks.Comment
-
Currently, I am running QC5 and QC6 on a dual Xeon 1GHz processor with 500MB ram. I just now set up a test workspace that charts one symbol on tick, 1 mnute and daily charts plus T&S plus has a Quote Sheet running with the first 490 NYSE symbols (alpha sort) monitoring Bid, BestBid, Net, BestAsk and Ask. The total CPU load including all CPU programs is averaging about 60%. If you peg the CPU with your workspace you either need more CPU power (dual should always be used) or you need to cut back your workspace.
~BobComment
-
jcgg,
Are you running any AOL programs or products on your PC?
'Tail.Comment
-
Hardtail,
No I am not. But it is all water under the bridge now. I went out and upgraded my system to a AMD X2 dual core system and everything hums along smoothly now.
Thanks for your interest though.Comment
-
jcgg,
I hope many read of your change and the problems that it solved. As I indicated in a prior response, I think a good dual core or dual CPU system will solve quite a number of the problems that many are having. I've been using dual CPU systems of one type or another for almost 10 years and once you have used a dual CPU system I think you would never switch back to a single CPU system again.
~BobComment
-
Really?
I've got 2 gigs of ram and am running a pentium 4 processor. Granted, the machine is from 04, but reallY? Does this program require a dual core processor to run well? My workspace has w, d, 233, 55 and 34 min charts, and I am monitoring about 150 ticker symbols.
If getting a new processor will fix pegging issues, I'll do it, but, seems like the program would run better than it does. Version 5 didn't peg it nearly as much.Comment
-
Matthew,
It is difficult to compare a workspace running on QC5 verses QC6 because QC5 was initially designed to work using the Continuum data feed and QC6 is using the eSignal data feed with some amount of inefficiency being introduced. Typically, the same workspace will require more CPU capability running QC6 than QC5. In my experience, a dual CPU system will perform better under high load conditions than a single CPU. If you would like me to run your workspace on one of my dual CPU systems just send the workspace to me and I will post my results.
~Bob
[email protected]Comment
-
Matthew,
What my friend Bob says is so, and I appreciate his enthusiasm regarding dual core/processor machines. A couple or three additional comments if I may.
First though, regarding your specific question....
"Does this program require a dual core processor to run well?"
Nope, not as an absolute requirement. Don't get me wrong, my next box will be dual core, but the essential factor is that my next box will have more horsepower to accommodate QCharts 6.0. There's no essential "bug", at least that I'm aware of, in QCharts that means it will run more correctly on a dual core(or processor) machine vs a single processor. That is, there's no programming flaw in QCharts that requires a dual machine. Regarding running QCharts, there are single processor machines with higher MIPS rates (Millions of Instructions Per Second) that will outperform dual core/processor machines that have a lesser MIPS capacity. There are folks I've talked with that have dual processor machines that are experiencing %cpu issues with 6.0. And again, my next machine will be dual core for several reasons, but more horsepower, a higher effective MIPS rate being the essential factor to accommodate QCharts 6.0.
As Bob notes, there are two factors at play regarding 6.0 incurring more %cpu vs 5.1 when running the same workspace.
1. QC6 gets it's data from the eSignal feed, as you know. There are more ticks/sec coming in, especially in the first and last hour of a market day. The QCharts continuum feed(QFeed) is on a tick-throttled pipe from Comstock...the so called IP2 feed. QCharts will have to do more processing (%cpu) to handle the additional #ticks streaming in moment to moment. The subsequent greater %cpu is to be expected, nothing wrong there.
2. QC6 has significant processing efficiency issues unrelated to an increased #ticks. So significant that these inefficiencies are, imho based on some observations, as significant in increasing %cpu as are the greater #ticks in the eSignal feed that cause an increased %cpu. I need to quantify which is most significant and by how much, but for reasons I won't detail here, it's difficult and I'm not sure it's possible to separate the two factors accurately. Anyhoo, it's for sure the 6.0 processing inefficiencies are significant. I have demonstrated that comparing 6.0 and 5.1 running the same #tics with 5.1 running on a not-throttled feed(the so called IP4 feed).
Post-106301 Posted: 10-18-2007 09:53 PM
I need to run another test on the latest version of 6.0. I'm up against the wall right now on some other stuff, but hope to get to that this week.
I'm running a 2.4Ghz P4 uniprocessor with plenty of RAM. I've cut my workspace to account for the greater #ticks in eSignal feed. In other words, I've constructed a workspace I'm OK with, that gives me a universe of data that's acceptable, and this workspace runs OK on the continuum IP4 feed which is roughly equivalent to eSignal feed regarding #ticks. So, if 6.0 didn't have processing inefficiencies, I would be good to go as is on my current box. However I'm not happy that a further cutting of the workspace is required to accommodate the processing inefficiencies that currently exist in 6.0. I'm not happy at the prospect of having to buy a box with more horsepower just to accommodate what I consider to be bugs (inefficiencies) in 6.0.
Jay,
Yes indeedy there is an expected and valid increase in %cpu with 6.0 due to increased #ticks. But sorry good buddy, I just can't let the processing inefficiency issue be glossed over, or subjugated to a general notion that the greater %cpu in 6.0 is mostly and primarily and for all practical purposes all about increased #ticks. Nothing would make me happier for it to be proven that my observations are in error, that I've missed something essential in my observations. But my only agenda is sharing actual factual reality, and I've been careful to compare apples to apples as much as possible.
LAMComment
-
Larry,
Bob is helping me out with some hands-on support.
I messed around with q6 all day, and became more and more frustrated. It consistently pegged my machine at around 90% usage.
I needed to see charts, so I used q5.1 again, and it ran flawlessly, running at around 7% cpu usage consistently.
My cpu is clean, with a p4 processor, and 2 gigs of ram.
I'm not viewing too many symbols at a time.
My conclusion (which is actually a hopeful hypothesis at this point) is that my download is glitched somewhere. I'm marinating this through in my head, and soliciting help (thanks again Bob) here.
To be sure...I'm not the only one with this problem.
Sure, things would always be better with more (ie a dual-core processor) but, I just can't see that being the problem with my particular situation.Comment
-
Matthew and Larry,
Matthew sent me his workspace and even though I am only running 2 CRT monitors and Matthew is running 4 flat screen monitors, it does not make sense that I was showing under 20% CPU use and Matthew was showing almost 80% CPU. That kind of percentage difference cannot be attributable to the dual core/single core difference.
Matthew, if Larry would not mind, maybe you could also send him your workspace and we could all run it on QC6 for a short time tomorrow and then compare differences.
~BobComment
-
Matthew,
Just saw Bob's post, and sure, please send me your workspace.
[email protected]
Sorry to dash your hopes regarding your comment....
"My conclusion (which is actually a hopeful hypothesis at this point) is that my download is glitched somewhere."
If that were so, that would mean fewer ticks coming in. And that would mean, by definition, a "reduction" in %cpu. I'm not saying that you don't have connectivity issues to the eSignal servers, that you're not having a connectivity related freeze in data delivery and then a catch-up surge that will overwhelm the cpu, but my cursory take is that's "probably" not the issue.
6.0 chews up more %cpu for the two factors noted previously.
1. More ticks/sec in the eSignal feed. This is to be expected.
2. Processing inefficiencies chewing up cpu cycles.
I can't run my workspace...the one that's already been slimmed down to account for the greater #ticks in eSignal feed...in 6.0, except for mid-day when the market trade rates are usually significantly lower than market open or going into the close.
LAMComment
-
A dual core is not necessary to run 6.0. Neither of my machines is a dual, and I have no problem running on either one. Granted 6.0 is consuming more processor availability, but should not impact your 5.1 workspace unless you are over your symbol limits. I run 6.0 right up until close and hold positions over night and take profits/losses the next day. From my point of view there is no difference between the middle hours and the last hours of trading as far as pulling down symbols to view. My 1.1ghz processor works nearly as well as my 3.2ghz and in reality, I actually do more trading on the 1.1 as I am at work at the close of trading. You may refer to some of my earlier posts to view the screen-shots of my platform.
'Tail.Comment
-
HardTail,
I never said or implied a dual processor was necessary but if a workspace is dragging the CPU to it's knees the dual CPU will help. IMO, you are giving people false impressions by indicating that you are running QC6 on a 1 GHz processor. Let's get some facts out here -
1. How many symbols are you monitoring in QC6?
2. Are these symbols high trade (over 10,000) or low trade (under 1,000) or ???
3. How many monitors are you running?
4. How many charts are you running and are the time frames "standard" or "non standard".
We were talking about Matthew's workspace with over 150 symbols and 5 charts with two charts having "non standard" time frames. Matthew was not able to run this workspace at all times on QC6 with a P4 processor.
Tell us more about your hardware and workspace.
~BobComment
Comment