AMD's SB750: Enabling Higher Phenom Overclocks?
by Gary Key on July 23, 2008 12:00 AM EST- Posted in
- CPUs
Final Words-
Actually, these are not our final words as we have significantly more testing to complete on newly arriving boards with the soon to be introduced SB750 South Bridge. So where do we start?
Let's see, AMD has introduced a new South Bridge that is certainly an upgrade to the long in the tooth SB600 and also addresses one feature flaw in the SB700. That feature flaw is RAID 5 and while we are not proponents of host controller RAID 5, early indications is that AMD's implementation is no worse than NVIDIA's or Intel's current solutions. Marketing can check that one off the feature list now. However, we are still disappointed in the PCIe 1.1 interface to the 790FX/780G and upcoming new NB chipsets.
AMD is probably correct that there is plenty of bandwidth available with the current solution, but we cannot help but wonder why they did not include a PCIe 2.0 interface given the amount of time this chipset has been under development. Also, the lack of a native LAN interface is perplexing, not from a performance viewpoint, but the fact that additional costs are incurred by the motherboard companies and ultimately users for the additional interface components. For a company that is specializing on reduced cost and highly integrated solutions, this one is perplexing. However, in the end, the performance of the SB750 rivals that of its competitors at this time, or comes close enough in current testing that we can overlook our. Speaking of performance, early numbers (just received final chipset drivers and late beta RAID drivers) indicate a slight improvement over SB700 but well within the error of margin in our storage test suite.
Of course, the big news is the new low level interface between the SB750 and the Phenom CPU. Advanced Clock Calibration is its name and overclocking is its game (a bad throwback to days gone by). In all seriousness, ACC works and depending on just how "substandard" your Phenom might be at overclocking, it works extremely well. We have several theories about what this interface does and how AMD is doing it, but without explicit information or conformation from AMD we will have to continue theorizing about it. Our testing continues for the new product launches next week, but we were satisfied enough with the current numbers to publish early results. What we are not satisfied with is the lack of information from AMD, but that is their prerogative.
Our test results confirm AMD's statements that a 100MHz to 300MHz improvement in Phenom core speeds are attainable with ACC. This is not a guarantee and some processors will require significantly more tuning than others to get up to speed, but overall our experiences to date with several processors are very positive. Our otherwise clock challenged 9850BE suddenly started acting like a Tour de France rider on steroids with ACC enabled. Our clock improvements ranged from 200MHz to 410MHz depending upon our settings. What was particularly interesting is the fact that our largest improvements occurred when we tried raising HT ref clock and increasing the multiplier at the same time.
Our 9850BE responded especially well in this case with ACC enabled, so much so that we wondered if maybe our original settings were completely off. They were not as it turns out after a few hundred reboots between setting changes on several boards. We also found it was much easier to sustain a high NB speed with ACC enabled and this lead to several significant improvements in memory performance. Finally, we found leaving the program on auto settings provided the quickest and easiest overclock improvements, at least with our processors. Utilizing AOD was much simpler than making constant changes in the BIOS, but in the end, neither one offered an improvement over the other one in performance.
We purchased a retail processor that seems to have been blessed by the processor Pope before it left the FAB and arrived in the office. As such, we only recognized a 100MHz improvement in core speed with ACC enabled on the Foxconn board. We were able to tune the SB600 equipped ASUS board to the point where the difference was only 50MHz at the top end. What we did not report on today since we are still verifying the test results, is that ACC did make a significant difference in NB speed with this processor in the midrange clocks from 3GHz to 3.2GHz with NB speeds ranging from 2520 to 2600 at these speeds. Compared to the Foxconn board with ACC turned off or the ASUS board, our maximum NB speeds hovered around 2200 to 2400 with slightly more CPU and NB Vid required.
AMD informed us that we should be able to run lower voltages and improved HT ref clock speeds with the Phenoms. We are currently testing these claims but at this point we do see lower voltages being required at a variety of clock speeds but have not noticed any true improvements in HT ref clocks, yet. As always, your mileage will vary depending upon your CPU and motherboard, but we believe most users will see some improvement, maybe not enough to spend additional money on upgrading the motherboard. That said, we feel like this technology will benefit the enthusiast and feel like it is probably a safe purchase at this time for those users who have to have the latest and greatest technology.
The complaints about the Phenom processor not clocking as well under Vista 64 when compared to Vista 32 are for the most part true. We had varying degrees of separation in the final core speed numbers depending upon settings, enough so that we would have to recommend XP Pro or Vista 32 over their 64-bit counterparts for those who need to squeeze the last ounce of performance from their systems. Once again, the differences varied greatly depending on the motherboard, processor, BIOS settings and if ACC was enabled or not. While the final system settings under Vista 64 were acceptable to us, we know for many this will not be the case. We hope to have some answers soon from AMD.
In the end, we know after a good 60 hours of testing, over 600 different settings, 400+ screen shots, and countless reboots that Advanced Clock Calibration works, yet we do not know anymore about the inner-workings of ACC at this point than SpongeBob SquarePants.
Facebook
Twitter
RSS
48 Comments
View All Comments
bbqchickenrobot - Monday, August 11, 2008 - link
This article claims the PCIe slots are 1.x when they are in fact 2.0 as stated in the manual, website and other sources.Just wanted to make everyone aware of this. There are four total PCIe 2.0 that can run 2x PCIe 2.0 cards at x16/x16 or 4x PCIe 2.0 cards at x8/x8/x8/x8
To get around this, get an X2 version of the Radeon HD 3870 or 4870 and then crossfire with another one or drop down to a standard 3870/4870
MikeODanyurs - Wednesday, August 6, 2008 - link
Anyone see AMD OverDrive 2.1.2 available for download yet? It's Wednesday the 6th, official release day of 790GX+SB750.CZroe - Monday, July 28, 2008 - link
"it's unclear whether or not whatever AMD is doing here can even work on K8 if they tried."Nope. I've been waiting for a board/BIOS maker to expose this chipset setting since the Athlon XP Mobile Barton CPUs first made me aware of it. My Mobile Barton Athlon XP was basically a completely unlocked bin-sorted Athlon XP 3200+ that was stable at lower voltages expected to be run at much-lower voltages and clock speeds.
With it, I discovered that there was clearly a chipset setting not exposed when my Mobile Barton failed to reach 200/400FSB in an nForce2 Ultra400 board (the Shuttle board inside my SN45G XPC... the FN45-whatever IIRC) while the same CPU could easily do 450+FSB in an Asus A7N8X-E Deluxe and an Abit NF7-S. The solution? Drop a thin wire inside the socket to make this completely unlocked CPU *DEFAULT* to 400FSB and it runs perfectly fine and can even go higher (the L12 trick/mod).
With the default FSB being the only difference (all settings/frequencies, including the FSB, were the same), it was very clear that it was related to a calibration setting that was automatically applied by the chipset based on the default FSB (100MHZ/200FSB for this unlocked CPU). Previously, I could only get to 333FSB before Memtest 86+ would start giving errors. There were some reports of even the overclocking boards (mentioned earlier) getting higher with the L12 trick and later many had BIOS updates and board revisions that seemingly integrated the trick. Regardless, the setting was still not exposed and was likely not even something the board/BIOS makers could expose (though they could spoof the default FSB I'd assume). It looks like AMD, having full control of chipset and board-level options here, has finally let the cat out of the bag.
Although the FSB has become the HT bus, the behavior is the same. Changing the calibration setting allows for higher clocks while actually becoming unstable at lower clocks. Ever accidentally erased the OSCAL calibration value on a PIC microcontroller and had to find a working value through trial and error? I'm guessing that this is why it was automatic before.
initialised - Sunday, July 27, 2008 - link
Leads me to suspect that ACC is a means of adjusting the substrate potential to provide a bias to the bottom of the SOI stack to control the threshold voltage of the transistors.SirYes - Friday, July 25, 2008 - link
To spread the word: http://ubuntu-virginia.ubuntuforums.org/showthread...| To be honest I find it hard to believe that they have a campaign
| against Linux. There would be nothing to gain and would have
| a zero fiscal gain.
|
| I think it is a bug you have found
After looking through the disassembled BIOS for the last several hours, rebooting it, and tweaking it more, I'd say this is very intentional, I've found redundant checks to make sure it's really running on Windows, regardless what the OS tells it it is, and then of course fatal errors that will kernel panic FreeBSD or Linux, scattered all over the place, even in the table path for Windows 9x, NT, 2000, XP, and Vista, and had to correct them (Well, at least divert them off into a segment of RAM I hope to god I'm sure about)
No, this looks extremely calculated, it's like they knew someone would probably go tearing it apart eventually and so tried to scatter landmines out so as to where you'd probably hit one eventually.
So if it is a mistake, or incompetence, then it's the most meticulous, targeted, and dare I say, anal retentive incompetence I've seen.
Will14 - Friday, July 25, 2008 - link
I've never had much faith in Windows OC tools.However if this one actually works and AMD can up their speeds a little then I may switch back for my next system.
I must say my 6400-8400(cheated got an e3110 with 20% discount when first released and they were all $280 or OOS for $170). Upgrade path was seamless and I feel bad going from 1950Pro to 8800GT 512 but I follow my wallet not the company I root for.
A cheap system built off cheap Black Box edition(OC'd easily with their tool) and 4870x2 would be pretty sweet come this December if the new cores are kicking and the utils work(a lot of ifs). Not that I need a new system though, but if one did. I don't see nehalems being affordable anytime soon(unless new AMD core rocks(another if)), although my mobo would support. I like the $200 mark for processors $100 for HDD's $200 for gpu's plus generally only needing 1-4 new parts for a yearly $400ish upgrade which is cheaper and allows better performance than the $1-2k some people build.
Zisyncmon - Friday, July 25, 2008 - link
Does anyone know when they will release more info? Gary? Any estimates of when to expect another article? Or will we have to be in suspense all weekend.geok1ng - Thursday, July 24, 2008 - link
I am really hoping that AMD can make some competition on CPU market as they did on the IGP (man i am really looking forward to the next IGP offer from AMD, the 780G is the king, the next chip will be a market consolidating move for sure)and VGA markets but i am not joking when i ask:Will a 3.1Ghz ultraoverclocked quad core Phenom beat a $80 E2180@3,1Ghz in gaming and common day tasks?
As i remember the C2Ds are 15%-25% faster clock per clock, and the AMD CPUs consume MORE power for the same tasks , because they take LONGER to do the same task of a C2D.
jslusser - Thursday, July 24, 2008 - link
So does this ACC on the SB750 do anything for vanilla Phenom's? Will it unlock the multiplier on the non-black Phenoms or just make them more stable with the HT and memory over clocks?I'd like to see where the 9150e/9350e can get to with ACC and the power usage, since the voltage, etc doesn't change. So many questions...
HazaroudsSmoker - Friday, July 25, 2008 - link
First, you can't unlock the multiplier on the non Black Edition Phenoms. They are locked at the factory during manufacturing and can not be altered.Sounds like there will be an advantage still to having the SB750 with Vanilla Phenoms.
Can't wait for Benchmarks.