Check out the latest article on the AT front page about the new ULi chipset and southbridge that provides for a real AGP slot in addition to PCI-Express.
In the comments, I noticed someone pointing out an ASRock board using this chipset that also had a slot for an "upgrade riser card" for Socket M2 in the future. Information and a forum discussion (including pictures, and now a review) can be found here (note that it seems they have changed the board's name from 939Dual-M2 to 939Dual-SATA2 - but the M2 upgradability is still there).
The highlights (for those who don't want to follow the links):
[*]Native AGP 8x
[*]PCI-Express x1 and x16 slots
[*]PCI-E based SATA2 controller
[*]'ASrock Future CPU Port for upgrading to M2 in the future' :Q
This board seems perfect for me, almost too good to be true. I'm sure I'm not the only one with a high-end AGP video card (6800GT in my case), wanting to move to an Athlon 64 system, but not necessarily wanting to replace the video card immediately. From the article here on AT, AGP performance is shown to be top-notch, so that's not a concern as it has been with some of the partial AGP solutions on other PCI-E boards (such as those that basically put the AGP slot on the PCI bus). Then of course the upgrade path for a nice PCI-E card is there later.
That alone would make a very compelling motherboard in my eyes, but the possibility of upgrading to Socket M2 in the future makes it even more so. As far as I can tell by doing a few quick searches, the upgrade cards (for example, a Socket 939 card for an older 754 motherboard) cost about $30...very inexpensive upgrade path. Also, their older motherboards with these upgrade cards seem to do well in benchmarks too.
I also like the fact that the board seems to have three PCI slots instead of the usual two - since my HDTV tuner and its DVI card take up two slots, and I'd then still have a slot for a sound card if I wanted one. I realize not everyone cares that much about them though.
As far as I can tell from the comments in the thread I linked to, it appears this board could be pretty close to release, maybe even within the next month. I have to admit that I never really considered a motherboard from ASRock before, or anything based on a ULi chipset, but the board looks to be just what I need, and the chipset appears to be solid from Anand's latest preview. The only downside is that it doesn't have a ton of crazy overclocking options and voltages like the DFI's do - but it would probably be sufficient for me. The CPU voltages appear to be enough to get a good overclock from a Venice or X2, and I'm running standard PC3200 RAM so I don't necessarily need more memory voltage anyway.
I'll be looking at this very closely when it appears, hopefully soon.
Your thoughts?
Edit: Feel free to discuss the chipset itself in here too and not necessarily the specific motherboard (and Socket M2 stuff) I mentioned... The author of the front page article mentioned in the comments that Abit and Gigabyte are working on boards based on this chipset too. :thumbsup:
Edit #2 (9/7/05): AT now has a review of the board on their front page.
The board is currently listed at Newegg for $68 (back in stock once again ), and at eWiz for $66.95, in stock (prices current as of 11/3).
Edit #3 (9/8/05): It appears that Jetway is working on a ULi M1695/1567 board as well, but very little info is currently available.
Also, 939Dual-SATA2 Bios 1.20 is available at ASRock's website - I'm not entirely sure whether shipping boards are already coming with this version or the older/original 1.10 version.
ASRock's product page for the motherboard, for reference.
Edit #4 (9/12/05): Users at the OCW Forums are reporting that the XP-120 does in fact work on the board. In addition, there is a new beta BIOS 1.21 available for download; users are reporting somewhat better overclocking results in general, but as always, install beta BIOS versions at your own risk. No higher voltages yet, as far as I can tell.
Edit #5 (9/18/05): It looks like there is now a working Vcore mod for the 939Dual-SATA2, tested up to 1.65V. While obviously it would be better to have more voltage options directly in the BIOS, at least we know the possibility is there now. This might be the last piece of the puzzle (for me anyway, but probably others as well).
Update 9/20: New (somewhat simpler) Vcore and Vdimm mods are here at RHCF. Also OCWorkbench has posted their own Vdimm mod and Vcore mod.
Edit #6 (9/28/05): New beta BIOS available, reported changes are mainly support for more processors, such as the new Socket 939 Opteron 144.
(10/20/05) BIOS 1.3 available, looks to be mostly support for more processors again. (10/21/05) There are reports that this new BIOS has some serious issues with voltage adjustments, including for users who have done the volt mods, so you may want to stick with an older one (1.2 maybe) for now.
Edit #7 (11/3/05): BIOS 1.4 is now available, looks to add only official support for the new Socket 939 Opterons. There are still a few reports that voltage adjustments are still messed up since version 1.2, so use caution if you decide to try this one out.
(11/9/05) New OCW Beta Bios 3 now available. Users report better results with 1T, as well as higher HTT speeds than the previous betas. This is based on the latest official BIOS 1.40, so it includes the latest support for all the CPU's listed in that version, including (I believe) the new Opterons. Also, it appears that the Rebel's Haven Vcore mod (1.65/1.70V after soldering a resistor, the first one I posted) might be causing some problems for some people. Remember to be very careful when doing any mod, and that your warranty will most certainly be voided. More on this as it develops.
(12/9/05) Official BIOS 1.50 now available. Adds official support for the FX-60, even though it's not out yet. No other changes that I've read about.
(1/21/06) OCW Beta BIOS 4 was released a short time ago. Appears to be based on the official BIOS 1.50, and possible advantages (or disadvantages) over other versions is unclear. I couldn't find much discussion on this BIOS even on the OCW Forums except for one relatively short thread. Seems that OCWB4 still has a ~274HTT lock, and people are still getting the best results out of OCWB1/2 (especially if they're limited to a 9x multiplier and need lots of HTT speed).
(2/7/06) New Official BIOS 1.60 now available (DOS/Windows). Apparently this BIOS "Add one item 'AGP P2P Deep Fifo' in BIOS setting."
(3/23/06) New ULi integrated drivers available, version 2.20 (FTP link). A few updates for the LAN, USB, and SATA/RAID drivers (though no word on what those updates actually are). Many of the included drivers have been tested and apparently work on the latest Vista build (according to the readme file), which is good news.
(4/12/06) New (Official) BIOS 1.70 now available (DOS/Windows), along with OCW Beta BIOS #6. The one major change listed is support for the AM2 CPU riser board (which I don't believe is even available yet). Also, several users are now reporting better results/successes with 1T command rate on the RAM with this BIOS. No real word on overclocking performance yet, except that the infamous "274MHz Lock" still seems to be in place.
I have read suggestions at the RHCF forum that setting CPU-NB Link Speed down to 200MHz mostly fixes the 274 lock problem even on newer offical BIOS releases, I've not yet been able to do much testing on it myself but I haven't been successful so far. Still, I thought I'd throw that info out there for anyone who might be interested in trying it.
(4/18/06) OCW Beta BIOS #7 now available, reportedly with no more 274MHz HTT limitation!
(4/30/06) OCW Beta BIOS #8 now available, based on the new Official BIOS 1.80 (DOS/Windows). The official 1.80 BIOS is notable because it's the first official ASRock BIOS to remove the 274 HTT lock (I'm running at 300x9 on 1.80 right now). So if you for whatever reason were reluctant to try out the beta BIOS releases, and wanted higher HTT speeds available to you, give 1.80 a try. I can't tell any obvious differences between 1.80 and OCWB8, although I've seen at least one post from someone claiming better compatibility with his RAM on OCWB8 than he saw on 1.80.
Oh, and both of these new releases, in addition to the Socket AM2 support that was introduced in Official BIOS 1.70, now have an update for the AM2 CnQ algorithm. Of course, you can't yet actually buy the riser card or any Socket AM2 CPU's...
(5/29/06) ASRock's download page now has BIOS 1.90 up; modifications are "for AM2 memory code". Also, OCW did a review of the AM2 riser card, and I believe they tested it on the 939Dual-SATA2. No word on when it'll be available at retail, though I imagine it must be soon. As long as it's priced inexpensively, it'll be a nice cheap option for those who are interested.
(6/4/06) The ASRock download page now lists BIOS 2.10; official changes are support for the AM2 3500+ and a bit of tweaking for 5900 series video cards. Also, I think the AM2 CPU riser board is starting to appear in online retailer inventories for those who may be interested...
(6/23/06) I've posted about it in the thread, but I might as well update this post for people looking for a quick summary - the Socket AM2 upgrade riser card is available at several places now, the cheapest of which being eWiz at $30.90.
(6/27/06) ASRock's download page now has BIOS 2.20: "Add new CPU support: AM2 Athlon 3800+ (ADD3800IAA5CU)(through AM2 CPU board)". Just in case anyone actually is using the AM2 board
(10/10/06) Yes, I'm still here and checking this thread from time to time. BIOS 2.30 is now out on ASRock's download page for this board (has been for a week or so actually); only change listed is "Patch nVidia 5950 AGP card.". I kinda doubt anyone's using one of those on this board, and they don't even mention what they patched. No obvious signs of improvements from looking at the thread over on OCWforums either. Just in case anyone really wants to play with something new.
In the comments, I noticed someone pointing out an ASRock board using this chipset that also had a slot for an "upgrade riser card" for Socket M2 in the future. Information and a forum discussion (including pictures, and now a review) can be found here (note that it seems they have changed the board's name from 939Dual-M2 to 939Dual-SATA2 - but the M2 upgradability is still there).
The highlights (for those who don't want to follow the links):
[*]Native AGP 8x
[*]PCI-Express x1 and x16 slots
[*]PCI-E based SATA2 controller
[*]'ASrock Future CPU Port for upgrading to M2 in the future' :Q
This board seems perfect for me, almost too good to be true. I'm sure I'm not the only one with a high-end AGP video card (6800GT in my case), wanting to move to an Athlon 64 system, but not necessarily wanting to replace the video card immediately. From the article here on AT, AGP performance is shown to be top-notch, so that's not a concern as it has been with some of the partial AGP solutions on other PCI-E boards (such as those that basically put the AGP slot on the PCI bus). Then of course the upgrade path for a nice PCI-E card is there later.
That alone would make a very compelling motherboard in my eyes, but the possibility of upgrading to Socket M2 in the future makes it even more so. As far as I can tell by doing a few quick searches, the upgrade cards (for example, a Socket 939 card for an older 754 motherboard) cost about $30...very inexpensive upgrade path. Also, their older motherboards with these upgrade cards seem to do well in benchmarks too.
I also like the fact that the board seems to have three PCI slots instead of the usual two - since my HDTV tuner and its DVI card take up two slots, and I'd then still have a slot for a sound card if I wanted one. I realize not everyone cares that much about them though.
As far as I can tell from the comments in the thread I linked to, it appears this board could be pretty close to release, maybe even within the next month. I have to admit that I never really considered a motherboard from ASRock before, or anything based on a ULi chipset, but the board looks to be just what I need, and the chipset appears to be solid from Anand's latest preview. The only downside is that it doesn't have a ton of crazy overclocking options and voltages like the DFI's do - but it would probably be sufficient for me. The CPU voltages appear to be enough to get a good overclock from a Venice or X2, and I'm running standard PC3200 RAM so I don't necessarily need more memory voltage anyway.
I'll be looking at this very closely when it appears, hopefully soon.
Your thoughts?
Edit: Feel free to discuss the chipset itself in here too and not necessarily the specific motherboard (and Socket M2 stuff) I mentioned... The author of the front page article mentioned in the comments that Abit and Gigabyte are working on boards based on this chipset too. :thumbsup:
Edit #2 (9/7/05): AT now has a review of the board on their front page.
The board is currently listed at Newegg for $68 (back in stock once again ), and at eWiz for $66.95, in stock (prices current as of 11/3).
Edit #3 (9/8/05): It appears that Jetway is working on a ULi M1695/1567 board as well, but very little info is currently available.
Also, 939Dual-SATA2 Bios 1.20 is available at ASRock's website - I'm not entirely sure whether shipping boards are already coming with this version or the older/original 1.10 version.
ASRock's product page for the motherboard, for reference.
Edit #4 (9/12/05): Users at the OCW Forums are reporting that the XP-120 does in fact work on the board. In addition, there is a new beta BIOS 1.21 available for download; users are reporting somewhat better overclocking results in general, but as always, install beta BIOS versions at your own risk. No higher voltages yet, as far as I can tell.
Edit #5 (9/18/05): It looks like there is now a working Vcore mod for the 939Dual-SATA2, tested up to 1.65V. While obviously it would be better to have more voltage options directly in the BIOS, at least we know the possibility is there now. This might be the last piece of the puzzle (for me anyway, but probably others as well).
Update 9/20: New (somewhat simpler) Vcore and Vdimm mods are here at RHCF. Also OCWorkbench has posted their own Vdimm mod and Vcore mod.
Edit #6 (9/28/05): New beta BIOS available, reported changes are mainly support for more processors, such as the new Socket 939 Opteron 144.
(10/20/05) BIOS 1.3 available, looks to be mostly support for more processors again. (10/21/05) There are reports that this new BIOS has some serious issues with voltage adjustments, including for users who have done the volt mods, so you may want to stick with an older one (1.2 maybe) for now.
Edit #7 (11/3/05): BIOS 1.4 is now available, looks to add only official support for the new Socket 939 Opterons. There are still a few reports that voltage adjustments are still messed up since version 1.2, so use caution if you decide to try this one out.
(11/9/05) New OCW Beta Bios 3 now available. Users report better results with 1T, as well as higher HTT speeds than the previous betas. This is based on the latest official BIOS 1.40, so it includes the latest support for all the CPU's listed in that version, including (I believe) the new Opterons. Also, it appears that the Rebel's Haven Vcore mod (1.65/1.70V after soldering a resistor, the first one I posted) might be causing some problems for some people. Remember to be very careful when doing any mod, and that your warranty will most certainly be voided. More on this as it develops.
(12/9/05) Official BIOS 1.50 now available. Adds official support for the FX-60, even though it's not out yet. No other changes that I've read about.
(1/21/06) OCW Beta BIOS 4 was released a short time ago. Appears to be based on the official BIOS 1.50, and possible advantages (or disadvantages) over other versions is unclear. I couldn't find much discussion on this BIOS even on the OCW Forums except for one relatively short thread. Seems that OCWB4 still has a ~274HTT lock, and people are still getting the best results out of OCWB1/2 (especially if they're limited to a 9x multiplier and need lots of HTT speed).
(2/7/06) New Official BIOS 1.60 now available (DOS/Windows). Apparently this BIOS "Add one item 'AGP P2P Deep Fifo' in BIOS setting."
(3/23/06) New ULi integrated drivers available, version 2.20 (FTP link). A few updates for the LAN, USB, and SATA/RAID drivers (though no word on what those updates actually are). Many of the included drivers have been tested and apparently work on the latest Vista build (according to the readme file), which is good news.
(4/12/06) New (Official) BIOS 1.70 now available (DOS/Windows), along with OCW Beta BIOS #6. The one major change listed is support for the AM2 CPU riser board (which I don't believe is even available yet). Also, several users are now reporting better results/successes with 1T command rate on the RAM with this BIOS. No real word on overclocking performance yet, except that the infamous "274MHz Lock" still seems to be in place.
I have read suggestions at the RHCF forum that setting CPU-NB Link Speed down to 200MHz mostly fixes the 274 lock problem even on newer offical BIOS releases, I've not yet been able to do much testing on it myself but I haven't been successful so far. Still, I thought I'd throw that info out there for anyone who might be interested in trying it.
(4/18/06) OCW Beta BIOS #7 now available, reportedly with no more 274MHz HTT limitation!
(4/30/06) OCW Beta BIOS #8 now available, based on the new Official BIOS 1.80 (DOS/Windows). The official 1.80 BIOS is notable because it's the first official ASRock BIOS to remove the 274 HTT lock (I'm running at 300x9 on 1.80 right now). So if you for whatever reason were reluctant to try out the beta BIOS releases, and wanted higher HTT speeds available to you, give 1.80 a try. I can't tell any obvious differences between 1.80 and OCWB8, although I've seen at least one post from someone claiming better compatibility with his RAM on OCWB8 than he saw on 1.80.
Oh, and both of these new releases, in addition to the Socket AM2 support that was introduced in Official BIOS 1.70, now have an update for the AM2 CnQ algorithm. Of course, you can't yet actually buy the riser card or any Socket AM2 CPU's...
(5/29/06) ASRock's download page now has BIOS 1.90 up; modifications are "for AM2 memory code". Also, OCW did a review of the AM2 riser card, and I believe they tested it on the 939Dual-SATA2. No word on when it'll be available at retail, though I imagine it must be soon. As long as it's priced inexpensively, it'll be a nice cheap option for those who are interested.
(6/4/06) The ASRock download page now lists BIOS 2.10; official changes are support for the AM2 3500+ and a bit of tweaking for 5900 series video cards. Also, I think the AM2 CPU riser board is starting to appear in online retailer inventories for those who may be interested...
(6/23/06) I've posted about it in the thread, but I might as well update this post for people looking for a quick summary - the Socket AM2 upgrade riser card is available at several places now, the cheapest of which being eWiz at $30.90.
(6/27/06) ASRock's download page now has BIOS 2.20: "Add new CPU support: AM2 Athlon 3800+ (ADD3800IAA5CU)(through AM2 CPU board)". Just in case anyone actually is using the AM2 board
(10/10/06) Yes, I'm still here and checking this thread from time to time. BIOS 2.30 is now out on ASRock's download page for this board (has been for a week or so actually); only change listed is "Patch nVidia 5950 AGP card.". I kinda doubt anyone's using one of those on this board, and they don't even mention what they patched. No obvious signs of improvements from looking at the thread over on OCWforums either. Just in case anyone really wants to play with something new.
- Thread closed due to its age. Feel free to start a new thread with any new issue regarding this board.
AnandTech Moderator
lopri