Does 790FX + SB750 = High-End Overclocking?

[beerpapasmurph]

I do believe Toms Hardware is showing a little bias here towards Intel. Come on, did you not even try to push this processor/MB combo. I have the exact setup you tested. ASUS board, HD4870, AMD 9950 BE and Im running at 3.23 Ghz with bumping up the Multiplier to 16X and a slight Frequency adjustment up to 203. BTW ACC is set at 4X. Was really hoping for some good numbers to compare, but instead we get this as a "review". Thanks for wasting my time and yes Intel fans boys your stuff is faster, but the only reason you can buy equipment at that price is because of AMD not just throwing in the towel.

 

[Pei-chen]

[citation][nom]jadedgamerx[/nom].........Not to mention they are willing to hold onto old platforms that are more than sufficient for doing low horsepower tasks at a low price with low power requirements while Intel moves from socket to socket taking the upgrading path away from it's customer's and forcing them to re-engineer their entire setup around new ram, motherboard and cpu all at once......[/citation]
Are you kidding? In the same time Intel moved from LGA775 to LGA1366, AMD moved from Socket 940 to 939 to AM2 to AM2+ to AM3 coming next year.

 

[hannibal]

Hard to say anything about those overclocking results, but if I am not wrong, there can be guite big difference in CPU's about how high they eventually can go.
To me this says that you can get anything between 2,8 MHz to 3,4 MHz. Depending on how lucky you are. One thing is that in Toms articles they seems to be guite conservative in their overclocking, but that is just a feeling...

 

[jadedgamerx]

[citation][nom]Pei-chen[/nom]Are you kidding? In the same time Intel moved from LGA775 to LGA1366, AMD moved from Socket 940 to 939 to AM2 to AM2+ to AM3 coming next year.[/citation]
Yes, but there is absolutely no backwards compatible versions of Intel's new chips for old motherboards. AMD does update it's chipset socket often, but they almost always provide an upgrade solution on the previous generation platform as well. Switching from Athalon 64 --> Phenom was a very easy transition for almost everyone using an AM2 based board. Generally all that was required was a BIOS update. Boo Hoo..

 

[Pei-chen]

[citation][nom]jameskangster[/nom]Would you rather have AMD go bankrupt and end up selling their assets to ______ (fill in the blank) or stay in business and use everything they know to continue to improve their strong points?[/citation]
AMD is already owned by Abu Dhabi.

 

[megamanx00]

[citation][nom]jameskangster[/nom]In my opinion, it's just really difficult for AMD to continue to compete effectively against Intel at this rate. Even while Intel is slashing their sales outlook, they are still well funded without having to rely on external investments and loans. Also, their new processor lines based on i7 are very promising with early positive performance reviews. Heck, their current Core 2 architecture isn't even close to their end of useful life cycle. Meanwhile, AMD HAS NOT produced any new performance-competitive CPU SKUs (even price-performance advantage has been losing ground), and they have been LOSING desktop/laptop market share. The only positive product line left for them is in the graphics market, and they actually have gained market share with 48XX series. I think they should focus most if not all of their talents and funding in their graphics technology, and phase out their processor line. Also, I know it's good to have a competitor to Intel, but at what cost to the AMD emplyees and investors? Would you rather have AMD go bankrupt and end up selling their assets to ______ (fill in the blank) or stay in business and use everything they know to continue to improve their strong points?[/citation]

That's why the 45nm Deneb has been moved up. It was originally going to be launched in January, but I think AMD has realized hey REALLY can't wait that long .

 

[zodiacfml]

3.1 Ghz overclock is average as though they're using stock coolers.

 

[Pei-chen]

[citation][nom]jadedgamerx[/nom]Yes, but there is absolutely no backwards compatible versions of Intel's new chips for old motherboards. AMD does update it's chipset socket often, but they almost always provide an upgrade solution on the previous generation platform as well. Switching from Athalon 64 --> Phenom was a very easy transition for almost everyone using an AM2 based board. Generally all that was required was a BIOS update. Boo Hoo..[/citation]
You could also drop a Q9550 into a 945 or G31 board that's three and half years old.

The problem with upgrading from X2 to Phenom is that even with the best Phenom, you still couldn't touch C2Q so why bother. I bought a 690G + X2 for my sister a year ago believing AMD's long term upgradeability crap. A year later, the fastest CPU I can upgrade to is 9950 @ 2.6GHz or 3.0GHz if I can overclock that far on non-SB750 board.

Had I bought Intel an E2000 or E4000 CPU, I could simply OC to 3.0GHz for now and pickup a Q9000 in half a year to a year and keep the same board for four years. Now I can even be sure if AMD will continue to support AM2 board seeing how the CPU overclock based on south bridge.

BTW, I forgot to add AMD’s Quad FX (AMD 4x4) to the list. Intel’s Skull Trail was a waste but AMD was the one that came up with the ridicules dual socket non-sense and convinced hundred to buy FX-70/72.

 

[hellwig]

Why does this article keep mentioning the fact that these boards only support up to 3 dual-slot GPUs? This is a limitation of the 7-slot ATX form factor, NOT the particular layout of the PCIe slots. If I recall, manufacturers were showcasing a new 10-slot ATX-based form factor earlier this year. Until we see that hit the market, stop blaming the board manufacturers for conforming to the laws of physics (i.e. no two objects can occupy the same space at the same time, thus 4 cards requiring 8 slots cannot fit on a 7-slot board or in a case with only 7 slots). I suppose it matters if you don't keep your PC in a case (as when testing it out), but I have to think 99% of us do use a case, and that those cases currently have a maximum of 7 slots.

 

[Crashman]

[citation][nom]neiroatopelcc[/nom]From where I'm standing it's the other way around really. It's the manufacturers not included that don't care about the readers. Those who sent in boards are those who care about us readers - and dfi appearently isn't one of them at this point. You can't blame toms for that really.[/citation]

You make a strong point, and its one that's been implied in several articles. You might also take the view that sites write to promote the market: Honesty keeps the readers coming, while new product info keeps the manufacturers supplying components. The ideal end result is that people buy more stuff, more frequently, and are satisfied with their purchase.

 

[Crashman]

[citation][nom]Malovane[/nom]About the low overclock, I have to agree with the person who said it was a poor OC attempt. I myself am running a 9850 BE processor at 3.4 ghz stable at 1.4 volts...on air! That's using AMD Overdrive on a 790GX Foxconn board. I've pushed it higher.. just don't like my room to become a sauna. Anandtech got the venerable 9600 BE to 3.2Ghz, and others have pushed the 9850's and 9950's to 3.6ghz on water with little difficulty. You should be able to hit 3.1ghz with almost no voltage increase.[/citation]

Every processor is different. There's also the possibility that this particular sample was heat sensitive...and the cooler may not have been effective enough.

 

[Crashman]

[citation][nom]Beerpapasmurph[/nom]I do believe Toms Hardware is showing a little bias here towards Intel. Come on, did you not even try to push this processor/MB combo.[/citation]

Every motherboard gets a day of CPU overclock testing and a day of RAM stability testing in addition to benchmarking. If Tom's spent more time than that, it would take months to do the larger comparisons such as the 11-way P45 shootout.

[citation][nom]hannibal[/nom]Hard to say anything about those owerclocking results, but if I am not wrong there can be guite big difference in CPU's how high the eventually can go.To me this says that you can get anything between 2,8 MHz to 3,4 MHz. Depending on how lucky you are. One thing is that in Toms articles they seems to be guite conservative in their overclocking, but that is just a feeling...[/citation]

These motherboard articles apply settings that the reviewer feels are safe enough for many months (years, hopefully) of use. What was more disappointing is that ACC didn't have a noticeable effect on maximum clock speed. Perhaps compatibility of ACC with AHCI wasn't really fixed, even though the settings appeared in the later AOD version?

[citation][nom]Megamanx00[/nom]That's why the 45nm Deneb has been moved up. It was originally going to be launched in January, but I think AMD has realized hey REALLY can't wait that long .[/citation]

We've been hearing exciting rumors from professional overclockers about Deneb, we can only hope that some of those rumors are later verified.

 

[Crashman]

[citation][nom]hellwig[/nom]Why does this article keep mentioning the fact that these boards only support up to 3 dual-slot GPUs? This is a limitation of the 7-slot ATX form factor, NOT the particular layout of the PCIe slots. If I recall, manufacturers were showcasing a new 10-slot ATX-based form factor earlier this year. Until we see that hit the market, stop blaming the board manufacturers for conforming to the laws of physics (i.e. no two objects can occupy the same space at the same time, thus 4 cards requiring 8 slots cannot fit on a 7-slot board or in a case with only 7 slots). I suppose it matters if you don't keep your PC in a case (as when testing it out), but I have to think 99% of us do use a case, and that those cases currently have a maximum of 7 slots.[/citation]
Ten-slot Lian-Li case:

http://www.newegg.com/Product/Product.aspx?Item=N82E16811112159

If the x16 cards are positioned in Slot 1, 3, 5, and 7, you can fit four dual-slot cards into one motherboard using an 8-slot or larger case.

 

[Malovane]

[citation][nom]Crashman[/nom]Every processor is different. There's also the possibility that this particular sample was heat sensitive...and the cooler may not have been effective enough.[/citation]

This is true, as I've seen Q6600's not clock past 3Ghz, which is probably just the bottom of the bin. However, 9950's were binned for high heat and voltage tolerance, so you would think it would be fine. Perhaps they just had inadequate cooling, as you say.

It's just strange that a site which touts it's overclocking competitions and teams can't get a 9950 close to the average OC (by other sites and many enthusiasts), then basically makes the conclusion that AMD CPU's suck and 750 southbridge offers little improvement.... when even sites that aren't usually friendly to AMD contradict that... and came out with their 750 southbridge reviews and overclock results months ago!

Did anyone on the team think to look at what other people were getting for overclocks, scratch their heads and say "hey, maybe we're screwing this up somehow?"....

 

[fadirocks]

here is something interesting and long waited by AMD
http://www.internetnews.com/hardware/article.php/3784716/AMD+Finds+Its+Way+To+Shanghai.htm
it's finally here YAY!!!

 

[jadedgamerx]

[citation][nom]fadirocks[/nom]here is something interesting and long waited by AMDhttp://www.internetnews.com/hardwa [...] anghai.htmit's finally here YAY!!![/citation]
Not goona lie, I'm stoked to build a Shanghai system and see how it performs. I'm not concerned with it beating Core 2's or Nehalem, I am concerned with how much faster than the current Phenom's it will be. I hope AMD can stay in the market, the last thing we need is a $10,000 extreme edition processor and $1,500 base processor from Intel to F*$& our budget's over completely.

 

[neiroatopelcc]

[citation][nom]Pei-chen[/nom]You could also drop a Q9550 into a 945 or G31 board that's three and half years old.[/citation]
G31 3½ years old? The bearlake stuff first became available to the public in may last year or so. the 945 may be over 3 years old (dunno) but it isn't really useful anyway since it's not a dualchannel supporting one.

I do agree with your overall point though.

And crashman have a talk with some of your collegues! They know how to put more quotes in the same post - and they also tend to appear less 'defensive' and more 'honest' in their replies. You could learn something from chris actaully. Think about it, don't argue about it.

 

[Crashman]

[citation][nom]neiroatopelcc[/nom]G31 3½ years old? The bearlake stuff first became available to the public in may last year or so. the 945 may be over 3 years old (dunno) but it isn't really useful anyway since it's not a dualchannel supporting one. I do agree with your overall point though. And crashman have a talk with some of your collegues! They know how to put more quotes in the same post - and they also tend to appear less 'defensive' and more 'honest' in their replies. You could learn something from chris actaully. Think about it, don't argue about it.[/citation]

If you want honest...I think there's still an AHCI bug in ACC, even though the later AOD software was supposed to have addressed this. Tom's Hardware uses AHCI in motherboard comparisons...and some boards such as the Jetway unit require it for the southbridge-suppported eSATA "port" to function properly.

ACC appeared somewhat effective at lower VCORE, etc, but with the lower settings these systems wouldn't break 3GHz. At higher settings ACC had little to no effect, but those higher settings were needed to breach 3GHz.

 

[justaguy]

In light of the i7 overclocking fiasco and now this article that was apparently posted under time pressures instead of being delayed to allow for a good motherboard to show up I'm starting to wonder about THG's overclocking skill and/or commitment.

Time pressure or not, the article's value is significantly diminished by the hardware employed. I guess it's somehow better to throw something at the wall rather than miss an artificial deadline.

 

[jadedgamerx]

[citation][nom]neiroatopelcc[/nom]G31 3½ years old? The bearlake stuff first became available to the public in may last year or so. the 945 may be over 3 years old (dunno) but it isn't really useful anyway since it's not a dualchannel supporting one. I do agree with your overall point though. And crashman have a talk with some of your collegues! They know how to put more quotes in the same post - and they also tend to appear less 'defensive' and more 'honest' in their replies. You could learn something from chris actaully. Think about it, don't argue about it.[/citation]
In response to both you and Pei-chen:
I guess I wasn't specific enough. AMD seams to be gearing itself towards enterprise systems and multimedia rigs with enough power for every day users but sipping less KWH on your power bill each month. Hence my support for the 780G/790G platforms. The embedded graphics are not going to be good enough for anything but the most non-enthusiast gamer, but they are perfect for running a quiet, cheap, efficient media streaming/backup/home theater/internet/email/server/firewall box(s). With the correct processor selection (anything in the 45W bin) they become a high value setup for the money. No extra video card necessary and just enough power to be responsive but nothing overkill. I have a system that consumes less than 80 watts under full load with 500 gig green line hard drive, in comparison to my Core 2 gaming setup which generally eats 700 watts under full load with an SLI setup. Granted 400 of those watts are being consumed by the graphics cards, that still means that the rest of the system is consuming roughly 200% more power than my 780G chipset completely. It's a niche market, but they're hitting it well. I don't see many people running to intel for a "cheap" low range solution, I do see people buying E2200's and clocking them up to 3.0Ghz for cheap gaming rigs.

 

[jamesl]

[citation][nom]Tropoc[/nom]and the 9950 is cheaper then the q6600 its cheaper then about every intel processor out there (even the dual cores such as e8400 etc).[/citation]

where do you shop?
the 8400 is $20 cheaper on newegg
the 7300 is $60 cheaper on newegg

 

[Pei-chen]

[citation][nom]jadedgamerx[/nom]In response to both you and Pei-chen:I guess I wasn't specific enough. AMD seams to be gearing itself towards enterprise systems and multimedia rigs with enough power for every day users but sipping less KWH on your power bill each month. Hence my support for the 780G/790G platforms. The embedded graphics are not going to be good enough for anything but the most non-enthusiast gamer, but they are perfect for running a quiet, cheap, efficient media streaming/backup/home theater/internet/email/server/firewall box(s). With the correct processor selection (anything in the 45W bin) they become a high value setup for the money. No extra video card necessary and just enough power to be responsive but nothing overkill. I have a system that consumes less than 80 watts under full load with 500 gig green line hard drive, in comparison to my Core 2 gaming setup which generally eats 700 watts under full load with an SLI setup. Granted 400 of those watts are being consumed by the graphics cards, that still means that the rest of the system is consuming roughly 200% more power than my 780G chipset completely. It's a niche market, but they're hitting it well. I don't see many people running to intel for a "cheap" low range solution, I do see people buying E2200's and clocking them up to 3.0Ghz for cheap gaming rigs.[/citation]
With Nvidia's 9300/9400 coming into the market and E5200 @ $82, AMD's niche are no longer exclusive. But I agree with you that AMD used to have an advantage in low powered office PC.

 

[randomizer]

So... we don't actually know for sure what ACC does?

 

[zodiacfml]

I have wanted to buy an Intel system but always end up with an AMD one since it is cheaper overall only at the cost of performance.
If i had the money, Intel's chips today provide more performance for the money and better power consumption.

 

[unclefester]

Does 790FX + SB750 = High-End Overclocking?

Pretty hard to draw any conclutions from this article.Sounded more like a board comparison.
They never mentioned what was done to achive they're over clock.
Multiplier?FSB?
They used the same memory in all three boards.Which may give one board an edge over another.
None of them were volted to reach high end over-clocking.
I guess we have to realize if they spent the man hours to truely get a high end over-clock even with just three boards, when they finished they would be two technologies behind