Hardware & Devices

Dual processor slower than single? Is it all relative ?

5-Apr-10 9:17

Vaclav_Sal wrote:
What is "USB bandwidth"? I run into this somewhere, I think it is in Properties of some USB ports..

that's how fast a usb port can send data. For a USB2 port it's theoretically 480Mbit/sec. In practice with overhead factored in it's only about 40 MByte/sec.

Vaclav_Sal wrote:
What about "the USB on front are different than in back"? Not too technical description, referring to desktop box.

Depends on the manufacturer, but in general they're the same. The most common exception would be a system old enough it only had USB1 ports and had USB2 devices added by an expansion card.

Vaclav_Sal wrote:
I got USB to IDE "converter" to run CD drive and it is noticeably slower that IDE. Is that “bandwidth” limitation?

Maybe. PATA is capable of 133MByte/sec a bit over 3x as fast as USB2. however a 48x CD drive is only capable of sending up to 9MByte/sec which shouldn't be enough to cause any problems. If your computer is so old it only has USB1 ports the drive would be limited to 1.5MByte/sec which is about equal to an 8x cd drive. This would be enough to cause noticeable slowdowns. XP would warn you that you're putting a 2.0 device in a 1.0 port and that it will be running slower. I don't recall if win2k did this or not.

Vaclav_Sal wrote:

NT “Device Manager “ identifies tons of USB controllers, how do identify them physically?
I mean where are they - on the motherboard or on PCI adapter ( BTW I took it off – got too many internal USB ports) or external D-link USB adapter?

If you look at your mobo/its manual will probably have the ports numbered. I'm not sure if you can get matching IDs in device manager or not. If you pulled the expansion card though all the ports should be on your mobo. AFAIK Using an external hub shouldn't change anything in device manager because everything on it will be connected through a single port on your mobo.

Vaclav_Sal wrote:
If I can change this “bandwidth” I need to know this.

You can't. It's a hardware limitation. The only thing you can do is if your mobo only has 1.0/1.1 ports is to add a 2.0 expansion card to have some faster ports for external drives.

3x12=36
2x12=24
1x12=12
0x12=18

Vaclav_5-Apr-10 5:40

Vaclav_

5-Apr-10 5:40

I was happily developing on my Windows 2000 1.6 GHz Pentium PC when my geeky son-in-law railroaded me into "new computer is the answer".
Now I got dual processor 3.0 GHz piece of flaky, undependable....
I did not expect much improvement in compiling time ( For what else would developer want faster machine?). I have read here, somewhere, that VC++ was not designed for dual processor.
What is frustrating that I for example cannot start several directories copies with single click as I used to do on the old single processor. I understand that actual copying speed cannot be improved since I still copy from one slow device to another (disks).
But the windows management seems to be sluggish.
Does that mean that Windows 2000 was not designed for dual processors also?
Just curious.
Please – any commentary on obsolesce of my OS and or VC++ are not necessary nor welcome.
Thanks for reading.
vaclav

Re: Dual processor slower than single? Is it all relative ?

Dan Neely5-Apr-10 5:56

Re: Dual processor slower than single? Is it all relative ?

5-Apr-10 5:56

Win2k pro does support dual-processor (and you dual core should like identical to two single socket CPUs windows). The biggest limitation is it doesn't understand hyper-threading; but that's only an issue in scheduling when you have 2 more more HT capable cores. However I know XP had separate HALs (low level components) for single and multi-processor systems and didn't have a supported method to switch between them without doing an OS reinstall. (Starting with vista, the performance hit on the small fraction of uni-processor systems was considered to be less of a concern than the dev effort of maintaining two code bases.)

XP allowed swapping the HAL via tools intended for driver devs. I've no idea if it would work in wink2 as well or not. I'd strongly recommend making a backup before trying because if it goes wrong your windows install will be fubar.

http://forums.nvidia.com/lofiversion/index.php?t44294.html[^]

Edit: No idea what's up with your UI problems.

3x12=36
2x12=24
1x12=12
0x12=18

Vaclav_5-Apr-10 6:07

Vaclav_

5-Apr-10 6:07

Thanks for reply.
I was actually hoping to see some impovements in regenerating RAID 5.
Every time OS ( actully the dynamic disk service to be technical) runs regeneration it just takes over the processing. Other than that I am content. Sort of!
Vaclav
PS I run Advanced server - just for the RAID 5!

Re: Dual processor slower than single? Is it all relative ?

Dan Neely5-Apr-10 7:08

Re: Dual processor slower than single? Is it all relative ?

5-Apr-10 7:08

If that's your only problem I'm 95% sure switching to a multi-processor HAL will serve as a fix since your second core will still be free while the raid is rebuilding.

3x12=36
2x12=24
1x12=12
0x12=18

Vaclav_5-Apr-10 7:15

Vaclav_

5-Apr-10 7:15

I'll try it as soon as it is back to normal! My VC++ was doing goofy things so I am rebulding it right now. Of coure RAID 5 is aslo regenerating for other reason.
Vaclav

Synchronizing data capture with video

softwaremonkey22-Mar-10 5:22

softwaremonkey

22-Mar-10 5:22

Hi,

I have a video camera connected to a PC based video recorder and I need to trigger a reading on an A/D card in the PC for each frame from the video camera (possibly synchronized to the frame sync of the video signal).

Is there perhaps something in the video capture driver that would allow me to do this? We haven't selected the capture card yet so if anyone has had some experience with a particular one I would be very interested.

Any suggestions would be welcome.

Thanks

Tony

Re: Synchronizing data capture with video

Jonathan Davies23-Mar-10 0:46

Jonathan Davies

23-Mar-10 0:46

softwaremonkey wrote:
Is there perhaps something in the video capture driver that would allow me to do this?

What sort of level are you considering working at? It's been 10 years since I did a similar thing at an SDK level but from the manual of the (that version now obsolete Bandit) frame grabber card I still have, I see it has programatic option to synchronize frame grab with an external trigger such as, perhaps, some signal associated with your A/D:

External Trigger Detection Methods
Description: Detection method values used by BD_GetExtrigDetection and
BD_SetExtrigDetection.
Values:
BD_EXTRIG_DETECT_NEGATIVE_EDGE: External trigger detected on falling
edge of trigger input.

Again at a programaming level I recently used a Firewire bus to connect to a camera and was able to program what to do when a frame was grabbed:

LRESULT CMainFrame::OnVideoGrabbed(UINT /*uMsg*/, WPARAM wParam, LPARAM lParam, BOOL& /*bHandled*/)
{
	// Sent by a child video window when a bitmap, grabbed from a video frame has been saved
	// to a temporary file and which can now be loaded ready for processing or otherwise dealt with.
	// WPARAM is a pointer to the file name.
	// LPARAM is a BOOL indicating wether the attempted grab and writing to file of the bitmap succeeded.
	//			If LPARAM is FALSE the filename is invalid.

	LPTSTR lpFileName = (TCHAR*)wParam;
	BOOL bGrabbed = (BOOL)lParam;

	if(!bGrabbed)
	{
		// Could do something sophisticated here like...

So the type if interation you want is possible by various mechanisms.

Re: Synchronizing data capture with video

softwaremonkey23-Mar-10 2:47

softwaremonkey

23-Mar-10 2:47

Thanks,

Thats the sort of information I was looking for. Both solutions appear quite workable but I guess that it all depends on what the frame grabbing board/software supports.

I have found it difficult to get this level of information before you spend your money (and these things aren't cheap). For example I have tried without success to get this type of information for an Advantech board.

Still, it's useful to know that I am not asking for the impossible. All I have to do is check the various web sites for SDK documentation.

Thanks again Thumbs Up | :thumbsup:

Tony

RCA -> USB -> C++ : how?

428820-Mar-10 0:31

4288

20-Mar-10 0:31

I have a wireless camera with RGB exit cable (PAL) and I was planning to control the video via C++ code

The idea was to use something like EasyCap adapter (RCA to USB with a driver CD) to let the camera play video in my notebook but the question is: how to control the video stream via C++? I don't think the included drivers could be controlled by code

Any idea on how to solve the problem?

---

Most Ram in a laptop

programmervb.netc++19-Mar-10 13:16

19-Mar-10 13:16

I need a laptop with 12GB or more of ram anybody know of one...
I cannot afford the dell m6500.
I need that much ram but I do not need any frills just horsepower any suggestions?

Humble Programmer

Re: Most Ram in a laptop

T21024-Apr-10 19:58

T2102

4-Apr-10 19:58

Don't know how much dell m6500 costs. This one comes with 6GB RAM and is upgradeable / customizable to 16 GB, which is the most recognizeable on windows 7 home premium OS.
http://www.shopping.hp.com/product/computers/notebooks/ENVY15_series/rts/3/computer_store/VM247UA%2523ABA

Re: Most Ram in a laptop

T21025-Apr-10 20:09

T2102

5-Apr-10 20:09

Here's a cheaper laptop where you can spend around $1,600+ for 8GB of RAM with LED screen.
http://www.avadirect.com/product_details_configurator.asp?PRID=16495
I personally would prefer a dual core cpu (core i7-620M) if I was only using the machine for non-database software development and web browsing.

New Machine

programmervb.netc++9-Mar-10 15:14

9-Mar-10 15:14

I am probably going to get a new machine because my boss is wanting to do our develop in VMWare. I am wanting to get the most bang for my buck. Does anyone have experience using dual core vs quad core processors for this type of dev. We are writing VB.NET apps and are fixing to move to Windows 7 64bit and Visual Studio 2010 64bit. I know I need a ton of memory, not sure how much I am thinking at least 8GB. I am also wondering about solid state drives I have read these actually decrease VS performance.

In summary which is better

2.0ghz quad OR 2.6ghz dual
Solid state OR 7200rpm sata... It has to be a laptop I have not seen any faster drives.

Windows7 will have nothing installed except essentials and VMWare withing VMWare we will run Windows7 with dev tools.

Any opinions thank you

BTW I have googled this to death and most info I find is dated and with the improvements to solid state etc that is why I was wanting to ask someone that might have actually used it.

Humble Programmer

Luc Pattyn9-Mar-10 15:35

9-Mar-10 15:35

I'd rather pay for more RAM (16GB, if doable) than for solid-state disk (RAM is cheaper than EEPROM).

And IIRC people are not overwhelmed by Visual Studio using multi-threading, so maybe the faster dual-core is the better choice.

Smile | :)

Luc Pattyn [Forum Guidelines] [Why QA sucks] [My Articles]

I only read code that is properly formatted, adding PRE tags is the easiest way to obtain that.

programmervb.netc++9-Mar-10 15:39

9-Mar-10 15:39

Are you referring to 2010 or 2008 VS?

Do you know if I will see a big difference in 64bit vs 32?

Humble Programmer

Luc Pattyn9-Mar-10 15:53

9-Mar-10 15:53

mostly VS2008 (the real VS2010 isn't available yet).

you can't use more than 4GB of RAM without Win64, and RAM is your biggest friend.

by itself, some people have claimed Win64 is faster than Win32 on the same amount of RAM, however I am not convinced I understand how that would be.

FWIW: I would be looking for an Intel Mobile i5 (latest chip technology = most activity for given dissipation budget; and hyperthreading, i.e. each core has two sets of registers, hence can run two threads intertwined).

Smile | :)

Luc Pattyn [Forum Guidelines] [Why QA sucks] [My Articles]

I only read code that is properly formatted, adding PRE tags is the easiest way to obtain that.

programmervb.netc++9-Mar-10 16:31

9-Mar-10 16:31

Any opinion on the SSD vs regular hard disk...

The things I have read have said there is no real improvement but most of that info is dated at least a year.

I also wonder when they are comparing these hard disk are they comparing a mid range SSD to a velociraptor ?

I cannot get a 15000rpm drive for my laptop as far as I know so would the SSD be better maybe.

Humble Programmer

Luc Pattyn9-Mar-10 16:38

9-Mar-10 16:38

DRAM memory is faster than most other types of solid-state memory; it sure outpaces EEPROM. The one faster RAM technology is SRAM, but they are "small" (CPU caches are made of them).

On a Win64 system I would never buy and install SSD for speed reasons (maybe for fast boot, reliability, power considerations, ...); I would always opt for lots of DRAM and rely on Windows file caching, and if and when not satisfied by that, install a RAM disk (haven't seen many of those lately).

Smile | :)

Luc Pattyn [Forum Guidelines] [Why QA sucks] [My Articles]

I only read code that is properly formatted, adding PRE tags is the easiest way to obtain that.

programmervb.netc++9-Mar-10 16:46

9-Mar-10 16:46

Thank you very much friend. You are very knowledgeable.

Humble Programmer

Dan Neely10-Mar-10 2:22

10-Mar-10 2:22

programmer_vb.net_c++ wrote:
Any opinion on the SSD vs regular hard disk...

The things I have read have said there is no real improvement but most of that info is dated at least a year.

I bought an SSD last fall. Windows boot time went from 33s to 14; after logging in the normal period of disk thrashing that left the desktop unusable was gone. I can start Opera with 40-70 saved tabs, outlook with two multi-gigabyte PST files; and all the trash that shows up in the tray at once without anything lagging on startup. It was the most visible performance upgrade I've had since going from a k5-100 (on a 486 mobo) to a k6-350 over a decade ago. Cool | :cool:

3x12=36
2x12=24
1x12=12
0x12=18

Luc Pattyn10-Mar-10 3:52

10-Mar-10 3:52

Sure, reading a lot of files once will be much faster, so boot speed-up must be impressive. However software development is quite a repetitive job, using the same files over and over, and writing some; hence I'd rather spend some money on more DRAM instead of an SSD, hoping to make all but the first access to each file a lot faster.

Smile | :)

Luc Pattyn [Forum Guidelines] [Why QA sucks] [My Articles]

I only read code that is properly formatted, adding PRE tags is the easiest way to obtain that.

Dan Neely10-Mar-10 5:24

10-Mar-10 5:24

If boot/app startup was the only benefit, you're right an SSD wouldn't add much, but since blocking IO is common in many apps almost everything becomes more responsive. Measuring system responsiveness isn't anywhere as easy as benching boot/app startup; but it's the responsiveness gains that gave me the real wow.

If you have to choose between enough dram to keep all your working data in memory or an SSD obviously the former is more important; but windows vista and later don't report IO operations as complete until they've been flushed from the cache to the disk and dropping the latency from each write from ~10ms to a few microseconds actually is noticeable in normal useage. EDIT: I didn't mention this in my original post because "more ram than I'll actually use" has IMO been implicit in any custom configured PC for the last few years. Only crappy retail configs skimp there any longer.

Unless you're CPU bound however I'd put an SSD as more useful than a faster CPU because it eliminates the most generally apparent performance bottleneck; when random IO trashes a magnetic HD.

About 4 years ago I blew about $6-700 on a hardware raid controller and 5 HDs to try and boost performance; outside of benchmarks I never noticed a difference. The responsiveness gain going from an i7@3.85ghz with a single HD to the same system with an SSD is almost as dramatic as the difference between the same system with an HD and my netbook with an HD.

3x12=36
2x12=24
1x12=12
0x12=18

Luc Pattyn10-Mar-10 5:46