Hardware MSI P45 Platinum

http://www.bit-tech.net/hardware/2008/06/20/msi-p45-platinum/1

We check out the latest P45 flavour from MSI to see if its new GreenPower Center, funky new heatpipes and DrMOS features, mean it's worth the consideration compared to the competition.

 

I think overclocking failed due those "DrMOS" IC's. Such a high current flowing through such a small pins - compared to usual MOSFET's. I'm really sceptical about their efficiency.

 

there are many small pins, but they work just fine. The board would simply be unstable if it couldnt provide the power and that isnt the case - the BIOS is just not up to standard for ocing that's all. Digital PWMs have very similar small connectors too.

 

If you compare boards based on P35 chipset - do they overclock better than P45 ones?(Gigabyte EP45 or this MSI P45Pt) I have seen rather an average oc's so far.

 

But we're talking very early BIOS' still - P35s have had a year to evolve. The P45 offers far more features because we're running out of life on the antiquated front side bus, but getting all those working together is a task-and-a-half.

The board is already on sale from this week, that's why we've reviewed it, but we'd still recommended waiting until next month at least to see what happens with regards to BIOS revisions.

 

I'll keep that in mind.
I think I know what sort of problems unmature BIOS can cause

 

Why does the MSI board have (or needs) a 12v connector above the top pci-e slot and the Gigabyte doe not?

 

My GA-X38-DS4 has the 4 pin molex connector for the PCI-E slots, but the newer revision, GA-EX38-DS4 does not. I have 2 HD3870's and I don't use the 4 pin molex.

 

Your reviewing style is, as always, superb. I know you've got the 'conclusion' part, but I wonder if you've thought of doing an additional bit that shows the plusses and negatives? The things that would make you want to buy the board and struggle on or what would really put you off, etc.

 

Thank you, Kenco

I know what you mean, but cutting something down into pluses and minus means people simply skip to the last page and read the bullet points and score. It's never sometimes simple enough to condense into a few words as well because there are lots of factors to balance.

 

Concerning USB, I believe you can boot to DOS on a drive and sys the USB drive without a problem. DOS floppy images are easy to find, and the key will stay bootable. You can use that key to then make other keys bootable the same way. All the USB keys I have are bootable so I've never seen it as a big deal, but maybe I'm an exception

Hopefully MSI will include the EFI shell in their EFI update and have an EFI flash utility. As long as they include the FAT file system driver, you can connect a any FAT device that the BIOS has a block driver for, see it under the shell and flash from it.

 

are other manafactures intending to make it upgradeable from bios to efi or is it only msi

 

It's not that fact that it's hard to do, but the process I went through was:

BIOS needs updating - no in-BIOS flash utility, so try an in-Windows one. That doesn't work in 64-bit Vista and then the online update didn't work for me so I'll boot from a floppy (easy to make, but a pain in the arse because I have to fish out a floppy drive + disks). Nope, because the BIOS file is too big, so I'll have to research and see how to make a USB drive bootable - this is not something Windows natively supports, not something people are generally aware of and I feel it's a facility MSI should provide on its install CD.

Comparatively, with an Asus or Gigabyte board I just throw the BIOS file on USB then use the in-BIOS flash stuff - job done, 2 minute job. Intuitive and exceptionally simple.

Woodstock - Don't know mate, only MSI have confirmed they are pushing it so far. You can use UEFI with any form of Windows, but to read the BIOS functions from Windows (temps, fan speeds, overclocking) you'll need Vista x64 SP1

 

I thought Intel had put EFI on some it's boards, but they don't really advertise if they do.


It is possible to tell is a system is running EFI if it has the pattern 0x5453595320494249 in memory. That is the signature on the system services table. The next dword is the revision number of the EFI spec being used and if it is valid, the board runs EFI. I was hoping there was a tool out there to do this already, but I didn't see one

I'm no ACPI expert, but I thought this was dependent on the ASL code, and the system firmware if it is needed to act as a middle man. Did MSI indicate this as a limitation of their firmware?

 

Intel and Microsoft confirmed this is a limitation of all Windows versions except for Vista 64 SP1. OSX can read UEFI afaik, and so can many *nix stuff.

Intel doesn't really advertise much about what they do, other than CPUs and a bit of chipsets tbh. They rely on pushing marketing budget to others to let them do it for them.

 

I have had this board and cannot get anything over 450FSB stable. I am wondering if there is a more serious issue with this board as if you look on MSI forums I have the exact same problem as you and another user on the MSI forum. I have mirrored the bios settings of somebody who has it 500FSB stable but keep getting BSOD when windows is loading where he is stable and we are using the same memory crucial ballistix PC28500 and he is using an E8500 andme an E8400. This same cpu is stable at 520FSB on the MSI P35 Neo2 board.

 

It's simply the BIOS bugs and MSI are working on it, they've come leaps and bounds in the last few weeks since we got it, so give it another month and it should keep getting better. We'll come back to it when UEFI is launched and see if MSI has a compelling board.

I've sat down at events in Austria and back in Taiwan before Computex with MSI and I've seen the board do quite good things, but under very specific conditions. The P35 boards took at least 6 weeks after launch to get really good too.

 

Have you seen the size of the MOSFET package pins/bumps on the Gigabyte and Asus board? They're about just as tiny. In such packages as QFN used by DrMOS, the pins are of a standard size as defined by the package specs (QFN, TO-252 etc). To make up for it, where high currents are to be carried, the load is spread over several pins - as is done in the case of these DrMOS devices. On top of that, the DrMOS' package uses large square pads under the package itself as further electrical conducting surfaces. As such, you should look at the total conductivity of the device, not individual pins - the DrMOS arguably has a contact surface area that is as much as if not greater than an equivalent discrete system.

The DrMOS (Driver + MOSFET) concept is actually a good step in the right direction. It was proposed by Intel to help tackle the design issues inherent in trying to provide for increasingly power-hungry CPUs while raising the efficiency/stability bar and keeping costs and design complexity down - essentially killing a few birds with one stone. By incorporating the driver circuit with the high and low side MOSFET, you essentially have a complete package which reduces component count and more importantly, reduces trace lengths between the normally-discreet components to almost nothing. This improves performance by reducing the effects of trace resistance/inductance/capacitance - resulting in higher efficiency and switching frequency - while ensuring that those components are perfectly matched to each other (especially driver-to-MOSFET).

The one major downside is that motherboard manufacturers will have less flexibility in choosing components - though that is probably more a commercial/logistic (e.g. long-term/preferred suppliers and/or the ability to use one driver for numerous designs, merely changing the MOSFETs to ones that offer better performance/lower cost as needed in different motherboard models) than an engineering issue.

The DrMOS devices used on the MSI board are Renesas R2J20602, which are capable of pumping out a sustained 40A per device (effectively, per phase) under PCB temperatures out to about 90°C. Very decent for its size, and with 5 phases the circuitry should be able to sustain about 200A - which ought to be plenty sufficient for most overclocking work. The issue seems very much more to do with a very raw BIOS and related issues. Not helping that is the remarkably poor attention given to methods by which the BIOS is updated.

Overall, great review, and as always, superb pics I wonder what PWM controller IC MSI uses.... out of curiosity

 

Broken link?

By the way, I noticed that at least one photo enlargement link appears to be broken: the close-up shot of the power circuitry feeding the North Bridge, at http://images.bit-tech.net/content_images/2008/06/msi-p45-platinum/nbp2-8.jpg

Looking at the naming convention used in the other photos in the article, I figured it should be nbp-8.jpg, and on trying that out it worked

 

I'll fish that out for you...

EDIT: Super uber high resolution of the whole board (5.75MB JPEG Image - 2621px x 2103px)