Atomic I/O letters column #98
Originally published 2009, in Atomic: Maximum Power ComputingReprinted here August 25, 2009 Last modified 16-Jan-2015.
I've heard good things about the Tseng ET6000
My interest was drawn to the recent Atomic feature on all (decent) modern graphics cards. I am interested in getting a Radeon 4870 to replace my aging GeForce 8800 GT.
Looking to the Far Cry 2 results, I was surprised that even the 9800GTX+ couldn't do FC2 that well at 1680*1050. Firing up the FC2 benchmark tool I set the test to the same settings as mentioned in the magazine and proceeded to run tests at 1680*1050 and 1028*1024 resolutions with all settings to very high and 2x AA. The results were very close to identical when comparing my machine to your test rig with your 8800GT. I turned to see what the specs on the test rig were and were blown away to find that it had a Nehalem i965 and other droolworthy components.
My system:
Pentium Dual-Core E5200 @ 3.3GHz (266*12.5)
2GB G.Skill RAM 5-5-5-15
Gigabyte stock 8800 GT
Gigabyte G31 mATX motherboard
This gets almost identical frame rates to the test rig with 8800 GT.
Am I dumb or can my system equal a rig with a CPU costing twice as much as my entire PC?
David
Answer:
I've no idea whether you're dumb or not. (I'm here all week. Tip your waitress!)
But yes, for this test your computer certainly could be expected to perform about as well as any computer with the same graphics card in it. The reason for this has been true for years, and was still true when this column first appeared in Atomic magazine six months ago (in case you were wondering why David was asking about some slightly stale cards), and will continue to be true as long as the current CPU/graphics-card relationship lasts.
When you run Far Cry 2 (or any other demanding recent FPS - Crysis, Fallout 3, et cetera) at a high resolution with lots of pretty-stuff turned on, the load on the graphics card increases a lot more than the load on the CPU. So if the graphics card isn't a real screamer - which a GeForce 8800 GT isn't, for these games - then it becomes the limiting factor. No matter how much CPU grunt you've got, the frame-rate won't improve, because the system is waiting for the graphics card.
This CPU-and-3D-card relationship has existed since the very first consumer 3D accelerators. Being aware of it can save you money, too. If your frame-rate in a new game sucks, run something that monitors CPU load (Task Manager will usually do), play the game a bit, then look at the CPU-utilisation graph. If at least one CPU core isn't up around 100% while you're playing, then the graphics card is what's holding you up.
If one CPU core is pegged at 100% while you're playing, though, then the most you can expect if you upgrade your graphics card is the opportunity to enjoy the same lousy frame rate in higher resolutions.
(You shouldn't expect most games to fully load more than one CPU core in almost any situation. There are now a few games that get real benefit from multiple CPUs, but there still tends to be one super-CPU-intensive main thread, plus a few others that all together can't fully occupy another core.)
Weird Windows almost works
I recently reinstalled Windows on my machine using a fairly highly customised Windows XP CD. I used tools like RyanVM and nLite to slipstream SP3 and hotfixes, removed components, tweaked settings and used Bashrat the Sneaky's driver packs to integrate hardware support.
The new install is great, with one exception - Advanced Power Management seems to no longer exist on this machine, meaning I can no longer suspend or hibernate my system. This has never been a problem before.
Gigabyte's power management software tells me my motherboard doesn't support APM, there are no APM drivers listed in the Device Manager, and the Add New Hardware Wizard doesn't find any un-installed devices or components. I have tried a couple of registry tweaks trying to get APM back, but to no avail.
I didn't remove any components related to power management when tweaking, so the only thing I can think of would be that some of the driverpacks I installed are conflicting or have slight incompatibilities (I told the driver-pack integrator to "KTD" - Keep The Drivers for later use in the Windows driver folders, if that makes a difference).
Does this sound likely? What else could it be?
Alex
Answer:
When you change lots of variables in a Windows install, as you have in this case, it's
entirely possible for all sorts of weird phenomena to occur.
In this case, though, my first stop would be the BIOS setup. It's been a long time (like, maybe eight years) since APM and its younger cousin ACPI (Advanced Configuration and Power Interface) were at all likely to cause any trouble, but your BIOS probably still has the option to turn power management off and do things the old way. Including, for instance, assigning IRQs to devices according to what slot they're in. Turn off ACPI/APM and install Windows and you'll see the symptoms you're describing.
If this is actually the case, you'll need to reinstall Windows over the top of itself - a "repair" install ought to do it - to install the power-management-aware Hardware Abstraction Layer (HAL) instead of the one you've got at the moment. All of your currently-installed software should survive this - it just changes the low-level stuff to match your PC's "new" features.
This leads me to my second guess as to what's going on, though. I'm not sure if it's actually possible to do this, but perhaps your oddball Windows install has managed to install the non-power-management HAL even though ACPI is turned on.
Boot-sector brain surgery
I just gave my PC a new motherboard, CPU and RAM, and upgraded the boot drive too because the old one was, um, old, and that's what you told me to do in "Grease and hard drive change".
I've set up a dual-boot XP/Vista system on this "reconditioned" PC (I was previously still running XP), and just got to the stage of plugging the 750Gb second hard drive of my old PC into the new motherboard. (I didn't plug it in at the start because I was paranoid about accidentally installing XP and/or Vista on it by mistake.)
But now I can't access the drive, in XP or Vista. Disk Management in both OSes gives me the option of "reactivating" it, but when I try I get a "the operation did not complete" error. And the only other option is to convert the disk into a "basic disk" (apparently I made it a "Dynamic Disk" in XP - don't remember why), and lose all the data.
I've got backups of most of the stuff on the drive, but there's about 700Gb of data there, which'd take a while to copy back onto it from all those DVD-Rs.
Is there some partition-management or other program that can let me re-access this disk? Please don't tell me I have to put my old motherboard and boot drive back in the machine...
Neil
Answer:
I haven't encountered this problem in Vista, but I've seen it in XP. To solve it, you
need to convert the disk back into a "Basic" volume, without losing data. But the official
way to do this conversion is to repartition and format, which will hose all your
data.
Fortunately, if you haven't done any of the special stuff that makes dynamic disks desirable, like software RAID or changing partition size, you actually can convert a dynamic to basic and keep the data.
The technique is as simple as it is terrifying: Just hand-edit the drive's boot sector!
To do this, you'll need a "disk editor" program of some sort (old-timers like me may still call them "sector editors"). My first port of call when I'm looking for a system utility like this is the freeware archive at Pricelessware, but in this case Microsoft themselves have an adequate little editor, "Disk Probe".
Disk Probe (dskprobe.exe) is part of a "Support Tools" package that's been around since Win2000; you can find it here. That's the WinXP SP2 version; I don't know whether it works in Vista. So do this in XP, just to be on the safe side.
Whatever program you use, you want to open the appropriate drive (remember that drive numbering starts at zero; Disk Management lists drives by number), go to the very first sector, and on the "01C0" line change the third byte from "42" to "07". The appropriate byte is highlighted in...
...this picture.
Save the edited sector to disk and reboot and hey presto, the disk should be Basic again and accessible in all current Windows flavours.
All usual disclaimers apply, here; it is very easy to mess your computer up badly by blundering around with a disk editor. Note also that the "Home" versions of Windows Vista, like the Home version of WinXP, don't support dynamic disks at all. Neither does Windows on a laptop, oddly enough; Microsoft figured that there was no point enabling dynamic disks on computers that can usually only accept one internal hard drive.
If you'd like a more detailed version of these instructions, check out this post on Ars Technica.
Square cells
I've got one of those MP3 players in the shape of a cassette, that you can put straight into a car cassette player. Mine's called a "Digisette AR-264", but it looks like the one you reviewed in 2005.
Where can I find batteries for this player?
Molly
There's no room for a AAA cell in a cassette-shaped MP3 player.
Answer:
This question is more widely relevant than you might at first think, because several
small MP3 players use these sorts of rechargeable cells, often referred to as a "gum
stick" battery. They've actually been around since the 1980s, when they were used in
super-slim Walkmen.
(The incredible smaller-than-the-cassette-it-played
WM-10 Walkman, however, ran
from an ordinary AA cell.)
Gum-stick cells are quite wide, but thinner than a AAA cell, and because they're rectangular they don't waste space inside small devices. They're part of the family of "prismatic" batteries, which in this context pretty much just means any battery that isn't a disc or cylinder - normal 9V batteries and big lantern batteries are prismatic. The gum-stick type never really took off, though, so it's not the easiest battery to find.
I'm pretty sure there's only one size of gum-stick battery, usually sold with the designation "14M", but I wouldn't bet my life on that. Any 1.2-volt gum-stick NiMH battery that physically fits in your MP3 player should work fine, though.
(The "1.2-volt" part means there's only one electrochemical cell in these batteries, so they're actually technically just a "cell", not a "battery". This matters when you're searching for a dealer, because they may have the product listed as a "gum stick cell" or "14M cell" or something, instead of any kind of "battery".)
