Old 11-15-2008, 03:28 PM   #1
Fabian
Human being with feelings
 
Fabian's Avatar
 
Join Date: Sep 2008
Location: Sweden
Posts: 7,416
Default benchmark, unexpected results

I bought a new laptop and decided to benchmark my comps DAW-abilities. The results I got was quite unexpected to me.

Four comps, all Dell, all XP pro SP3;
Optiplex GX60, P4 CPU 1.7GHz (upgrade), 2GB 400MHz DDR, 40GB Maxtor 7200 IDE
M20 Precision, Pentium M 760 2GHz, 1.5GB 533MHz DDR2, 250GB Hitachi Travelstar 7200 SATA
Vostro 1510, Intel Core 2 DUO T5870 2GHz, 3GB 667MHz DDR2, 250GB WD Scorpio Blue 5400 SATA
Vostro 1510, Intel Core 2 DUO T5870 2GHz, 3GB 667MHz DDR2, 250GB WD Scorpio Black 7200SATA

The M20 and the Vostros are laptops, the GX60 an old stationary that's been upgraded with memory and CPU (no overclocking, though). The two Vostros are actually the same machine with changed HD. On the two Vostro's I had to disabled both the Dell WLAN and the ACPI battery controller (more on this and in another thread). Note that the M20 had the WLAN enabled and downloads running over it!

On the GX60 and the Vostro Black were installed the DuX audio optimized XP (see http://forum.cockos.com/showthread.php?t=15923, links to download here http://forum.cockos.com/showpost.php...&postcount=476). Apparently this works wonders on comps, and it really did on the GX60. Before installing the DuXP, the GX60 could hardly manage any Reaper project without snapping, crackling and popping.

All comps were tested with asio4all, 512 samples buffer. All test were run with Brad Sucks' Making me Nervous (great song, I really like it!) I let Reaper run in full screen mode playing the song, and I watched the performance meter (ctrl+alt+P).

Results:
GX60 - CPU avg 70%, FX ~30%
M20 - CPU avg 34%, FX ~22%
Vostro Blue - CPU avg 30%, FX ~16%
Vostro Black - CPU avg 30%, FX ~16%

I find two things surprising here. the first is the remarkable ability of the M20. With WLAN and downloads running, it manages to get quite close to the dual core Vostros (I'll check it with WLAN disabled once downloads are doen). The second, and maybe most remarkable thing is that the two Vostros come out equal. The slipstreamed XP supplied by DuX seems to have no effect whatsoever! The Vostro Blue ran the Dell installed XP sp3, and a slower HD (5400rpm, avg access time 12ms vs 7200rpm, avg access time 4ms) and yet it managed almost identical performance compared to the Vostro Black. Was the more expensive HD and DuX' hard work for nothing? I don't know, for the GX60 the DuXP sure made a difference, but I don't understand the results for the two Vostros. Any insightful comments are welcome.
Fabian is offline   Reply With Quote
Old 11-16-2008, 12:34 AM   #2
Ollie
Super Moderator (no feelings)
 
Ollie's Avatar
 
Join Date: Dec 2007
Location: On or near a dike
Posts: 9,834
Default

Some remarks:

1. If you're not streaming enough tracks to reach the bandwith or access time limits of an internal HDD, it simpy does not affect CPU load measurements at all . Brad Sucks is certainly not the right thing to test these things.

2. A stripped down OS (=less services running, less startup programs, less active processes) mostly frees up RAM, the impact on CPU load is veeery low. Same as above, you will notice a difference only when the system is at its limits on RAM and near 100% CPU load.

3. For the reasons mentioned above, benchmarking at low CPU loads is not saying much. Using as much resources as possible is separating the wheat from the chaff very fast. Three questions are interesting in this concern:

- How many plugins of some sort can I fit into the CPU?
- How many tracks can I record and playback at the same time?
- How much ASIO buffer do I have to maintain to get problem-free performance at 100% CPU load?

The first is saying something about the power you get for a mixing scenario (when latency is not important), the second is self-explaining. The third is important when you are e.g. using many VSTi (especially on CPU-wise slightly challenged laptops) and tracking stuff while or after mixing. That's also where the overall performance of the whole machine is on trial and where the tweaks kick in - it's the point where otherwise harmless services and processes may start to interfere and where the quality of the computer design shows up as well.
Some computers go all the way to 100% CPU at 256 samples buffer glitch-free and still responsive and others (with stronger CPUs, same OS+optimizations) are crapping out at 80%, needing 384 or 512 samples to become usable again and this would be even worse if the tweaks were not applied.

4. The Reaper performance meter is only reading the load within Reaper, not the overall system load. It is cool to compare plugin performance but using this for performance comparisons of the overall system is not recommended. Use Taskmanager instead.

Last edited by Ollie; 11-16-2008 at 12:39 AM.
Ollie is offline   Reply With Quote
Old 11-16-2008, 01:37 AM   #3
Fabian
Human being with feelings
 
Fabian's Avatar
 
Join Date: Sep 2008
Location: Sweden
Posts: 7,416
Default

Yes, you're right of course. I was thinking that the Brad Sucks tune supplied with Reaper was not stressful enough, but that was the best I had in terms of a generally available benchmark; others might want to test and compare their set-ups.

Now, wouldn't a set of Reaper benchmark projects to stress test comps be of interest. Obviously, someone more computer knowledgabale than me should put together such a set of benchmarks :-).

Thanks
Fabian is offline   Reply With Quote
Old 11-16-2008, 02:32 AM   #4
Ollie
Super Moderator (no feelings)
 
Ollie's Avatar
 
Join Date: Dec 2007
Location: On or near a dike
Posts: 9,834
Default

Creating a single "standardized" test project is not that easy, it must be scalable to be used on a vast variety of computers from old P3 machines to quadcores. For example, the first of the following projects will overchallenge a slower "Pentium Dualcore" (Yonah CPU) already. Many factors affect the results and therefore the interpretation of them is not always easy.

There are 2 projects floating around here:

1. "AllReaper" (http://forum.cockos.com/showpost.php...1&postcount=20)

It will load your Vostros to ~60-70% CPU (it's probably too CPU-heavy for the M20). You can duplicate audio tracks to put your disks under load as well, but without duplicating the FX on them (especially the ReaVerbs raise the CPU load fast) and create your own test project this way, in order to compare the two Vostros. Remove 2 of the 3 Reaverbs to make it fit your M20 and Optiplex.

2. This test http://forum.cockos.com/showthread.php?t=4503

It will merely max out your CPU fast, but there are quite some results to compare yours with which may be interesting.

Both will not use enough of the 3 GB of RAM in the Vostros, you'll need something like a few instances of a RAM eating VSTi for that, like a sampler (maybe ReaSamplomatic5000) with some fat samples. But even with the RAM and CPU maxed out, there will only be minor differences between the two Vostros, if any.

Quote:
Originally Posted by Fabian View Post
On the two Vostro's I had to disabled both the Dell WLAN and the ACPI battery controller (more on this and in another thread).
If you have questions on that, by coincidence I was forced to do quite some research on the new Vostros. But I'm afraid you won't like the answer. So if they do what you want, don't ask.
Ollie is offline   Reply With Quote
Old 11-16-2008, 03:24 AM   #5
Fabian
Human being with feelings
 
Fabian's Avatar
 
Join Date: Sep 2008
Location: Sweden
Posts: 7,416
Default

OK, now we're getting somewhere. Now I realise I didn't do my homework, I never searched the forum for "performance test" or similar before doing my own benchmarking. Sorry. Only excuse is I had enough to think about around my own four comps, replacing HDs and all...

I'll try those benchmarks and return with some info. Hopefully it is of value to someone else but me.

As for your Vostro findings, Steindork, please tell me, either here or by PM. Here are some of my own findings (the "other thread" promised in my first post).

Vostro 1510, latency, measured by dpclatency.exe
------------------------------------------------
Code:
HW switch	|  on	| off	| off	|  on	| off	|  on	|  on	|  off	|
Dell WLAN	|  on	|  on	|  on	|  on	| off	| off	| off	|  off	|
ACPI		|  on	|  on	| off	| off	| off	| off	|  on	|   on	|
1394		|  --	|  on	| off	| off	| --	| --	| off	|  --	|
NIC		|  --	|  on	| off	| off	| --	| --	| off	|  --	|

avg		| 210	| 345	| 345	| 210	| 80	| 80	| 80	|  60	|
peak		| 28000	| 28500	| 415	| 25000	| 130	| 97	| 2800	| 2000	|
on/off is set in device manager, except for HW switch

66 sec apart comes spikes from the WLAN, ~25ms
30 sec apart comes spikes from the ACPI, ~2ms

EDIT: A thought on assessing performance from within Reapers Performance meter contra the Task Manager: Surely the "outside environment" (from Reapers point-of-view) would only add sort of a DC level? The environment does what it does whether Reaper is running or not, right? So reading the performance on Reapers own meter would suffice?

Last edited by Fabian; 11-16-2008 at 03:28 AM.
Fabian is offline   Reply With Quote
Old 11-16-2008, 05:53 AM   #6
Ollie
Super Moderator (no feelings)
 
Ollie's Avatar
 
Join Date: Dec 2007
Location: On or near a dike
Posts: 9,834
Default

Ok. Here we go:

Your observations and measurements are typical and exactly what I (and others) got with DPClat.exe. I spent ages experimenting with configurations and searching for hints on Google until I found out what's really going on. Here's the explanation for the behavior:

The new Vostros (amongst other current Dell laptop series) have this problem with exeedingly high DPC latencies due to a design flaw. Everything that causes a DPC is taking ~10x longer than usual for that. This is the reason why you get real dropouts (instead of merely clicks) while the WLAN is on and why you have to deactivate the "Microsoft ACPI-Compliant Control Method Battery" driver. Measuring the battery capacity is causing DPCs which are multiplied by 10 too. BTW, charging the battery causes >3000µs DPC spikes as well, so you better take it out when recording important stuff (it's better anyway to store them at a cool place charged to 40-60%). All sorts of very normal system events are causing unusual DPC spikes, like e.g. a system notification ("Ding"), even making the taskbar pop up repeatedly (when set to "auto-hide") makes for additional 50µs DPC latency. The basic (or "average") latency is accordingly 10x (at least) higher than normal too. I think you know how unusual all that is since you have enough Dell computers that don't behave like that.

This sounds worse than it actually is since - as you already found out - you can get rid of the spikes and make it work with low latency audio, although with a pretty unusual amount of device/driver deactivations (instead of simply flicking the WLAN switch like on other laptops). DPC latency is somewhat depending on CPU load and so the multiplied DPC latencies of all sorts of standard system drivers may still deteriorate the realtime performance a bit at high loads, my Vostro could not keep the 256 samples up to 100% but it could be much worse.

But this is also the reason why you can't listen to music on a media player or a webradio stream without glitches while the WLAN is active and the reason why the performance of the WLAN itself is affected by its own DPC latency. The official (Dell forums) explanation that the whole issue is caused by the WLAN is plain wrong and the workaround "turn off the WLAN" quite unacceptable.

At the end of a 2 months period of procrastination, alledgedly lost emails, no return calls/mails at all and sending my argumentation with convincing screenshots to the Dell EMEA boss, they had to admit that my assumptions and conclusions are correct and they told me that just of all the new Precision workstations have the same problem. I'm pretty sure that at least some Inspirons are affected by the same glitch as well (identical DPC behavior). I was also told that Dell will not try to fix this "due to a lack of interest by customers" (very funny... almost all current Dell laptops have a quite similar problem, although with a different root maybe). So you have to live with the not very convenient driver deactivation/reactivation thing. (Tip: Make yourself a new shortcut on the desktop and call "DEVMGMT.MSC" directly from there and/or try hardware profiles).

Sorry, that's the story. There's nothing we can do about it. After a serious case of "Dell hell", I could return mine for a refund since it was coming with various severe build quality blunders anyway.

P.S. If you experience troubles with ASIO4ALL, install the old Realtek driver version R1.61. It's the only one that really works with the combination of ASIO4ALL and the ALC268 in the Vostro. But since the ALC268 has no way to monitor the input signal directly (grrr...), you may want to use an external interface anyway.
Ollie is offline   Reply With Quote
Old 11-16-2008, 06:01 AM   #7
Ollie
Super Moderator (no feelings)
 
Ollie's Avatar
 
Join Date: Dec 2007
Location: On or near a dike
Posts: 9,834
Default

Oh... missed your edit:

I don't think so, there may be "outside" stuff I don't know of being invoked by Reaper and CPU consuming devices like USB/FW disks or interfaces, probably the load caused by the ASIO driver and so on are not reflected in the internal performance meter but they are clearly a part of the whole system "DAW". So the Taskmanager is the recommended device to watch the overall load. What is "DC level"?
Ollie is offline   Reply With Quote
Old 11-16-2008, 12:30 PM   #8
Fabian
Human being with feelings
 
Fabian's Avatar
 
Join Date: Sep 2008
Location: Sweden
Posts: 7,416
Default

Thanks for the posting Steindork. Explains a lot. I will try out the slipstreamed DuXP and disable the WLAN etc and see what gives. I was thinking of using the Vostro only for DAWing, but maybe I'll double boot, one XP for DAW and another for... well, other stuff...

With "DC level" I mean that the oustide environment is static with respect to Reapers work, and so merely adds a fixed offset to the measured load, but you clarified that too; that's not the case, among other things since Reapers performance meter does not take the workload of the asio driver into account. Thanks.

Any comments about this: http://www.avielsegal.com/index.html (under the link "DAW optimization"). That guy surely managed some serious XP tweaking, but is it worth it all?

Thanks.
Fabian is offline   Reply With Quote
Old 11-16-2008, 05:01 PM   #9
Ollie
Super Moderator (no feelings)
 
Ollie's Avatar
 
Join Date: Dec 2007
Location: On or near a dike
Posts: 9,834
Default

Holy cow! That's quite some text... The problem is that we're at the border between belief and facts here - skimming over the hard to read lines there, some of all those super tweaks have been proven doing nothing (especially for DAW performance) or sometimes maybe even doing the opposite. IMO all those generic guides are of limited use only because some of the stuff depends very much on your system - e.g. "turning off font antialiasing / Clearview" has never done anything on my systems (besides causing sore eyes) but it might actually do something on some weird sort of onboard GPU from the stone age with a faulty driver and a dead rat lying on the chip.

No, I personally wouldn't go *that* far (again) if I'm not totally despaired. I think current systems (look at his system description) with their ample power basically run just fine without going all the way to cripple the system to DAW only operation. It's nice if you don't *have to* reboot to get some files over the LAN or fire up Reaper to lay down an idea quickly that came while writing a letter. Unless the NIC driver is spoiling the fun that is or you're trying to run Reaper on a 512MB RAM system.

But some of that stuff found in most other tweaking guides is known to be pretty essential like turning off indexing, getting rid of unnecessary run entries and services. There are some overlaps between DAW tweaking and "system hardening" (closing ports by disabling services for security reasons) and that's where turning off stuff makes the most sense IMHO. This way I end up running 17 processes (including 3 from antivirus) and ~15 services (if the computer is in the LAN) as well.

My personal criteria are "no need to change buffers up to 100% CPU load", "idle means idle and not >0-1%" and "event log stays clean" and that can be achieved without turning off even the useful stuff. YMMV as always and since you have 2 pretty identical computers, you can do quite meaningful comparisons yourself.

*Disclaimer*
This text reflects my personal views and not "the truth", everything is IMO, IMHO, and AFAIK kthxbai.
Ollie is offline   Reply With Quote
Old 11-16-2008, 05:40 PM   #10
bullshark
Human being with feelings
 
bullshark's Avatar
 
Join Date: Jan 2006
Location: traîne mes guêtres en Québec...
Posts: 5,390
Default

Quote:
Originally Posted by Steindork View Post
Holy cow! That's quite some text...
Sh#t, you can say that again! I have nightmares about going through all of that, only to find I've gained 0.0037% efficiency on my machine, but my network is unavailable, I can't scan or print and my desktop is permanently set to a picture of my mother-in-law.
bullshark is offline   Reply With Quote
Old 11-17-2008, 03:51 PM   #11
Fabian
Human being with feelings
 
Fabian's Avatar
 
Join Date: Sep 2008
Location: Sweden
Posts: 7,416
Default The saga continues

FWIW, here are some new benchmarking I did of the Vostros. I stressed the machines much more this time, running the AllReaper project pointed to by Steindork.

First. Performance measured by TaskManager did not differ much from performance measured by Reapers Performance Meter (CtrAltP). Insignificantly, I would say, which would mean that asio4all was no bottleneck, at 512 samples, 13ms buffer. No other apps were running (except Explorer and services, of course).

Second. I first ran the project as is, AllReaper Orig, 11 trx, 33 fx. Then, to stress things, I duplicated every track, so I got AllReaper 2x, 22 trx, 62 fx (4 fx were on the master track).

Third. Before each test I let the system idle for two minutes, then I took readings, then I let Reaper play the project and I took the measures from Reapers Performance Meter.

The comp specs you have in my first post.

Code:
		 AllReaper Orig	|  AllReaper 2x	|
                  Idle	| Play	| Idle	| Play	|
		+---------------+---------------+
Vostro Blue	| 48%	| 54%	| 75%	| 90%	|
Vostro Black	| 48%	| 55%	| 75%	| 90%	|
Again, identical CPU performance. BUT, with double the tracks the Vostro Blue emitted some snap, crackle and pops, not much, like a moderately clean vinly record, but it was there. Not so with the Vostro Black. So it seems that the increased HD performance and/or the slipstreamed DuXP really made a difference.

But now for a new strange thing. When I had tested the AllReaper 2x on the Vostro Black, I loaded a project of my own, not nearly as loaded wit FX's as the Allreaper 2x, but with equal number of tracks. Then the Vostro Black started emitting serious snap/crack/pops! I increased the asio buffer to 1024 samples, and everything ran smoothly. Then I changed back to 512 samples, and to my surprise everything ran smoothly also then. I'm not sure if this was a HW or SW glitch, but something was surley not right... Maybe the Vostro won't be able to match even my amateurish needs...
Fabian is offline   Reply With Quote
Old 11-17-2008, 09:39 PM   #12
Ollie
Super Moderator (no feelings)
 
Ollie's Avatar
 
Join Date: Dec 2007
Location: On or near a dike
Posts: 9,834
Default

Cool! Your T5870 has definitely a tad more muscles than the T5670 in my Vostro. Not noticeable in most other applications but Reaper (probably most DAW software) seems to translate even small differences in FSB, L2 cache size and core speed into noticeable more/less plugins.

Quote:
Originally Posted by Fabian View Post
Performance measured by TaskManager did not differ much from performance measured by Reapers Performance Meter (CtrAltP). Insignificantly, I would say, which would mean that asio4all was no bottleneck, at 512 samples, 13ms buffer.
512 samples are a quite relaxed setting which puts a low load on the CPU. The load is rising (maybe exponentially) if you reduce that. Many soft synth players will prefer less than 10ms (256 samples), if you use amp sims the whole round trip latency comes into effect and you want 128 or even 64 samples (but ASIO4ALL won't do that at higher CPU loads)

Quote:
Originally Posted by Fabian View Post
So it seems that the increased HD performance and/or the slipstreamed DuXP really made a difference.
Even though the Scorpios may not be the top performers of their class, they both certainly belch out 22 tracks playback (at standard sampling rate/bit depths) happily and feel a bit bored while doing this. The effect of a reduced workload due to Dux' XP comes out a tad stronger on the Vostros probably (due to the DPC thing I wrote above) but anyway I'm sure the lighter OS is responsible for the difference.

Quote:
Originally Posted by Fabian View Post
But now for a new strange thing. When I had tested the AllReaper 2x on the Vostro Black, I loaded a project of my own, not nearly as loaded wit FX's as the Allreaper 2x, but with equal number of tracks. Then the Vostro Black started emitting serious snap/crack/pops! I increased the asio buffer to 1024 samples, and everything ran smoothly. Then I changed back to 512 samples, and to my surprise everything ran smoothly also then. I'm not sure if this was a HW or SW glitch, but something was surley not right... Maybe the Vostro won't be able to match even my amateurish needs...
Don't worry. I had that on a few occasions as well. Don't forget that a 50 cent onboard chip with its WDM driver wrapped in ASIO4ALL is certainly something different than a real and solid native ASIO driver.
Ollie is offline   Reply With Quote
Old 11-18-2008, 02:56 PM   #13
Fabian
Human being with feelings
 
Fabian's Avatar
 
Join Date: Sep 2008
Location: Sweden
Posts: 7,416
Default

The decision to upgrade to a Scorpio Black was based on this http://www.tomshardware.com/reviews/...rive,2006.html.

As for the RealTek driver R1.61, I'm still searching. Any ideas where to find one? RealTek.com doesn't seem to carry it any longer, and those I've found elsewhere only claim to support ALC260 and/or ALC888. Same thing?

EDIT: Found this http://www.soft-go.com/download_driv...-R1.61_49.html for ALC260, 2-channel. Will try...

EDIT: Finally, I hope, this http://forums.driverguide.com/showpo...53&postcount=5

Last edited by Fabian; 11-18-2008 at 03:16 PM.
Fabian is offline   Reply With Quote
Old 11-19-2008, 05:17 PM   #14
Ollie
Super Moderator (no feelings)
 
Ollie's Avatar
 
Join Date: Dec 2007
Location: On or near a dike
Posts: 9,834
Default

The Realtek drivers are for all chips IIRC. Any R1.61 will work with yours.
Ollie is offline   Reply With Quote
Old 12-14-2008, 01:18 PM   #15
Fabian
Human being with feelings
 
Fabian's Avatar
 
Join Date: Sep 2008
Location: Sweden
Posts: 7,416
Default

Quote:
Originally Posted by Steindork View Post
The Realtek drivers are for all chips IIRC. Any R1.61 will work with yours.
I must thank you for the tip, Steindork. R1.61 works much better than the newer version my Vostro came pre-installed with. Much better...
__________________
// MVHMF
I never always did the right thing, but all I did wasn't wrong...
Fabian is offline   Reply With Quote
Old 12-17-2008, 12:23 AM   #16
Ollie
Super Moderator (no feelings)
 
Ollie's Avatar
 
Join Date: Dec 2007
Location: On or near a dike
Posts: 9,834
Default

Don't thank me, thank Michael Tippach (the author of ASIO4ALL) who helped me out with this in the middle of the night after I pulled my hair out with my Vostro for 2 days. He's a really cool guy!
Ollie is offline   Reply With Quote
Old 05-18-2009, 07:59 PM   #17
MasterHouse
Human being with feelings
 
Join Date: May 2009
Posts: 2
Default Specific instructions

Fabian,

Unfortunately I've only found your post today, after struggling with the same laptop that you have, the infamous Vostro 1510...

Can you be more specific about the steps you took to tweek your laptop to be an usable DAW, please?

Where to disable the "Microsoft ACPI-Compliant Control Method Battery" driver, for example?

Thanks a lot for any help you can possibly give me.

Amaury
MasterHouse is offline   Reply With Quote
Old 05-19-2009, 01:24 PM   #18
Fabian
Human being with feelings
 
Fabian's Avatar
 
Join Date: Sep 2008
Location: Sweden
Posts: 7,416
Default

Master,
I did these things.

First, I installed DuX' audio optimized XP (http://forum.cockos.com/showthread.p...udio+optimised). Drivers you get from Dell, only take the ones absolutely necessary. If you are going to use the built-in audio be sure to install the R1.61 driver linked to above. I do not really recommend using it, but still I do use it sometimes, and I have not disabled it.

Then, I disabled the ACPI. This you do in Control Panel > System > Hardware > Device Manager > Batteries, right-click and choose Disable. I also took the battery out, as pointed out by Steindork above.

Then I also disabled the wireless in the BIOS. During boot up, press F2, then under Advanced, disable On Board LAN Control and WLAN Control

I also fiddled with the services, (lots on this here http://www.blackviper.com/WinXP/servicecfg.htm). You change the Startup type of the services under Control Panel > Administrative Tools > Services.

I have these on Automatic:
Event Log
HD Input Service
Logical Disk Manager
Plug and Play
Remote Procedure Calll (RPC)
System Event Notification
Windows Management Instrumentation.

These are Disabled:
Background Intelligent Transfer Service
Fast User Switching Compatibility
Print Spooler
Remote Desktop Help Session Manager
Routing and Remote Access
Shell Hardware Detection
Themes
Wireless Zero Configuration

The rest are on Manual

In addition, I am very careful what I install on the machine. Apps that install CPU-hogging services are forbidden. Basically this means only installing "portable" apps, programs that can be "installed" by simply copying them to a directory. This includes SumatraPDF (fro reading the Reaper manual and ReaMix), Waveosaur, Textpad (needs installation but is OK), IrfanView, Anvil Studio (also needs installation but is ok).

With this I get down to 256 samples buffer with Asio4all (yes, you need to install that to, of course), that's about the best I can get.

HTH.
__________________
// MVHMF
I never always did the right thing, but all I did wasn't wrong...
Fabian is offline   Reply With Quote
Old 05-19-2009, 02:57 PM   #19
MasterHouse
Human being with feelings
 
Join Date: May 2009
Posts: 2
Default It worked!

Fabian,

Thanks a lot for your help! Now, finally I'm working.

All the best to you.

Amaury
MasterHouse is offline   Reply With Quote
Old 05-19-2009, 09:52 PM   #20
Fabian
Human being with feelings
 
Fabian's Avatar
 
Join Date: Sep 2008
Location: Sweden
Posts: 7,416
Default

Quote:
Originally Posted by MasterHouse View Post
Thanks a lot for your help! Now, finally I'm working.
Great. Have fun.
__________________
// MVHMF
I never always did the right thing, but all I did wasn't wrong...
Fabian is offline   Reply With Quote
Old 03-16-2010, 02:19 PM   #21
Fabian
Human being with feelings
 
Fabian's Avatar
 
Join Date: Sep 2008
Location: Sweden
Posts: 7,416
Default

Resurrecting my own thread here, sorry...

New Dell laptop, a Precision M4400 with Intel 2 Core Duo T9600 2.8GHz, 2x2GB DDR2 RAM, 6M L2 Cache, 1066MHz FSB, it came with W7 pro x64. After some hassle (see http://forum.cockos.com/showthread.php?t=52147 and http://forum.cockos.com/showthread.php?t=52814) I managed to double boot it with DuX XP (http://forum.cockos.com/showthread.php?t=30162) and W7x86 on a WD Scorpio Blue 160GB, 5400rpm disk. With nothing better to do, I benchmarked this machine...

Code:
       | CPU% 128 spls | FX Chain% | RAM MB |
------ +---------------+-----------+--------+
XP x1  |      49       |    29     |   64   |
   x2  |      73       |    58     |  108   |
-------+---------------+-----------+--------+
       |               |           |        | CPU% 64 spls | FX Chain% | RAM MB |
-------+---------------+-----------+--------+--------------+-----------+--------+
W7 x1  |      34       |    30     |   74   |    44        |   34      |  96    |
   x2  |      74       |    59     |  120   +--------------+-----------+--------+
-------+---------------+-----------+--------+
I ran cAPSLOCK's Allreaper project (http://forum.cockos.com/showpost.php...1&postcount=20 and see above) both in original (that's x1) and with duplicated tracks (x2). I let it play looped for 30 minutes and then when it was at 4:00 minutes I took a reading from Reapers performance meter (negligible difference with task manager). I used the ASIO4ALL driver (audio interface: Lexicon Lambda). I had wireless on in both cases, and W7 tweaked with "adjust for performance" and "background services"; XP was already tweaked expertly by DuX. XP had ACPI battery control off. DPC looked good in both cases. W7 played AllReaper x2 quite good even at 64 samples (as low as ASIO4ALL would go) buffer, but there was the occasional stutter which would make it useless when recording, so I regard that as not useful. XP couldn't play at 64 samples at all, only cracks and pops and stutter. At x3, neither install would play usefully...

One thing noted... this, http://forum.cockos.com/showthread.php?t=48372, the core behavior looks much nicer on XP, and I'm talking exactly the same HW, I double-boot from the same HD.

BTW, all this is with IRRT, I will try AHCI shortly... and I'll put in a 7200 disk...
__________________
// MVHMF
I never always did the right thing, but all I did wasn't wrong...
Fabian is offline   Reply With Quote
Reply

Thread Tools
Display Modes

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off

Forum Jump


All times are GMT -7. The time now is 03:22 AM.


Powered by vBulletin® Version 3.8.11
Copyright ©2000 - 2024, vBulletin Solutions Inc.