Monday, May 14, 2012

Gigabyte GA-890GPA-UDH3 built-in ATI HD4290 video vs Asus nVidia GTX 560 Ti

About a year ago, I acquired a Gigabyte GA-890GPA-UD3H motherboard. It has a built-in video card with one DVI-D digital port, and one VGA analog port. I never used the video card and always had it disabled. I was using a separate discrete video card from nVidia, previously a 9800 GT, and now an nVidia GTX 60 Ti from Asus .

I am now running a triple-monitor setup, with a pair of HP LP3065 30" 2560x1600, and a Gateway FHD2401 24" 1920x1200, fitted with a custom mount to run in pivot (vertical) mode. Since the GTX 560 only has 2 DVI-D ports, I decided to enable the internal ATI motherboard video card to get a 3rd port. Unfortunately, I ran into several issues.

  1. I connected both the large HP LP3065 monitors to the much faster nVidia GTX 560 Ti video card. These monitors are setup horizontally on my left. I wanted the boot display to be the leftmost HP monitor. In the Gigabyte BIOS, under "Advanced BIOS features" / "Init display first", I selected "PEG" in order to select the nVidia card as my boot video card. It booted fine to the correct display, but unfortunately, when the OS loaded, it turned out that the internal ATI video card had been disabled !

    This happened despite the fact that I had "Advanced BIOS features"/ "IGX Configuration" / "Internal Graphics Mode" set to "UMA" . I tried the other values, "SidePort" and "UMA + SidePort". No go. The only way to have the internal video card enabled was to select that card as the boot display. This didn't sit well with me, since I had a pivot monitor attached to it, and thus all the BIOS bootup messages would appear vertically ! This is clearly a Gigabyte BIOS bug. The only way I was able to solve it was to physically switch the monitor connections between the two video cards. I connected the leftmost HP LP3065 to the internal ATI card on the motherboard, and then the second HP LP3065 and Gateway FHD2401 to the nVidia GTX 560 . Then I was able to have the boot display on the leftmost HP LP3065, and run all 3 monitors with both GPUs.
  2. After solving the first problem, when Windows 7 loaded, it recognized both GPUs and 3 monitors perfectly. But I started getting strange messages that "Windows has detected your computer's performance is slow". This was not actually the case. But the messages were incessant, every 30 seconds or so. Windows was inviting me to disable Aero and switch to Basic. Even when I switched the Windows setting to "optimize for quality", the messages continued. I did about an hour's worth of google searches, and nothing I found helped. I concluded that this was a software conflict between Windows 7 x64, the nVidia 296.10 drivers, and the ATI Catalyst 12.3 drivers.

    The only way I was able to solve this was to disable the motherboard's ATI GPU. I had an old 8600GT video card lying around, and put in the second PCI-E x16 slot . I used it only to drive the smaller Gateway FHD2401 display. Now that both video cards are nVidia, everything works fine, without any pesky bogus performance messages.

Tuesday, May 1, 2012

USB composite device vs Gigabyte motherboards keyboard power-up

The two desktop in my home office feature recent Gigabyte motherboards, specifically models GA-990FXA-UD3 rev 1.1 and GA-890GPA-UD3H rev 2.1 .

One notable "feature" of these boards is that they only have a single PS/2 port for keyboard or mouse, but not both. This is a stupid design. If you use PS/2, you normally want to use it for both keyboard and mouse.

The omission of the second PS/2 port has forced me to purchase adapters to convert the PS/2 mouse signal to USB, more specifically the PS/2 mouse signal from my Iogear GCS84B KVM switch.

Simple passive adapters did not work. There was no mouse signal on the computer. I had to purchase smart active adapters. I literally tried 10 different models at Fry's Electronics and Micro Center before I found that was compatible with the mouse signal from the KVM switch, a noname one when I got the GA-890GPA-UD3H motherboard a year ago, and a Belkin model when I got the GA-990FXA-UD3 motherboard more recently.

These smart adapters actually take a pair of PS/2 signals for both keyboard and mouse, and turn it into a single USB signal. The computer sees it as a USB composite device.

I figured I would try to use it as designed, and connect both the keyboard and mouse to the adapter.

I immediately noticed a problem : I could no longer power the machine through the keyboard.

The BIOS on the motherboard normally allows you to power the system through the keyboard or mouse. You can select any key, or a password, or mouse motion or click. I choose a simple 1-character password option.

The feature just does not work when using the PS/2 to USB adapter . I tried all the possible options with mouse and keyboard. I set all the wake-up options to "Enabled" in the power management menu. But nothing worked, except pressing the power button on the front. This is OK for one of my systems that is close to my chair, but not OK for the other one.

This is a bit puzzling, because the BIOS otherwise supports the keyboard on this adapter just fine for purposes of entering into the BIOS setup menu.

I switched the keyboard connections back to the PS/2 port on both motherboards, leaving only the PS/2 mouse on the adapter, and powering up the machine through the keyboard worked fine again.

The lesson here is that 15-year old USB technology still has not caught up to good old PS/2. It could be argued that this is a problem with the motherboard implementation of USB power-on. But there are likely many other motherboards with the same problem.

Yamaha KX61 MIDI keyboard vs Windows 7 sleep mode

The Yamaha KX61 is a very nice USB-powered MIDI controller. I have owned a pair of them for about 3 years.

While trying to reorganize the dozens of USB cables in my home office, I found out the following problem : if the KX61 is hooked up directly to my PC's GA-990FXA-UD3 motherboard via USB, the PC is unable to stay asleep. More precisely, the PC will go to sleep, then wake right back up 2 seconds later. I found out that the KX61 was the problem through trial and errors by unplugging USB cables one at a time.

The Windows powercfg -lastwake did not show any reason for waking up.

The fix was to connect KX61 through a USB hub instead of directly to the motherboard. In this case, the computer was able to remain asleep. I'm not sure if this is a software or hardware problem. Probably a combination of both.