Information

Did you know you could become an X-bit labs fan on Facebook or follow us on Twitter?



Discussion

Discussion on Article:
New Overclocking Star: AMD Phenom II X4 920 Review

Started by: biffzinker | Date 02/03/09 12:52:55 PM
Comments: 14 | Last Comment:  02/11/09 03:51:27 PM

Expand all threads | Collapse all threads

[1-8]

1. 
The reason ACC doesn't work in the Phenom II X4 is ACC is supposedly included on the cpu, no need for the 750 south bridge.
0 0 [Posted by: biffzinker  | Date: 02/03/09 12:52:55 PM]
Reply
- collapse thread

 
Here is AMD's explanation on ACC:
Put as briefly as I can, things learned through developing ACC with 65nm Phenom were baked into our new 45nm Phenom II silicon. So, turning on ACC while testing the 45nm Phenom II CPU will not provide an OC increase. You can just as well leave ACC off for Phenom II OC testing. Since, the "go-fast" things we learned from ACC (and those CPU parameter adjustments) were factored into 45nm, the benefit unlocked previously by ACC in 65nm silicon is already being realized without having to use the ACC feature separately. Of course AMD continues to look for new and innovative ways to push our CPU performance… so we'll continue to experiment. Just wanted to set proper expectations for anyone thinking ACC should be unlocking additional headroom but aren't seeing it.
0 0 [Posted by: Gavric  | Date: 02/03/09 01:21:21 PM]
Reply
2. 
Great article, thanks.
0 0 [Posted by: blzd  | Date: 02/03/09 03:12:55 PM]
Reply
3. 
I become less and less convinced of the value of SYSmark 2007 with every passing day, particularly as an indication of general performance. It seems to favour high performance with fairly large (~4MB) working sets, to the exclusion of all else. This sharply defined requirement does not reflect a common desktop use case, and even the games do not show such extreme behaviour despite their being one of the principal consumer application types that can show benefits from extra performance of this sort. This can probably be explained by the games being real applications that exhibit a high locality of access within smaller working subsets, as indeed most real applications do.

It is important to remember that Enhanced Intel SpeedStep is implemented in such a way that during overclocking it doesn't change the processor core voltage, but still lowers their frequency under low workload.

I understand that this is more characteristic of ASUS's implementation than Intel's. Although I don't consider it an especially serious problem, it would perhaps have been better to point this out or otherwise to have used a motherboard from a different vendor for the power consumption comparisons.
0 0 [Posted by: MTX  | Date: 02/03/09 04:00:42 PM]
Reply
- collapse thread

 
Thank you for meaty comment

As for SYSmark, I am looking for better complex benchmark now. But I cannot find anything better than It or PCMark.

As for ASUS and SpeedStep. It's true that ASUS implementation is not good. And I will try to move to Gigabyte platform soon.
0 0 [Posted by: Gavric  | Date: 02/04/09 04:59:57 AM]
Reply
 
As I see it, the value (or supposed value) of SYSmark comes mainly from the fact that it tests actual consumer applications, albeit not in a very realistic way. I haven't used PCMark, but by reading its literature, one gets the impression that it offers little beyond what you already have in your existing selection of tests--and the same goes largely for SPEC CPU CINT2006. As a result, the only thing that comes to mind as a possibly viable alternative to SYSmark is the Phoronix suite (http://www.phoronix-test-suite.com/). Again, I haven't used this personally, so I can't offer a specific recommendation, but it looks like it contains tests which could be helpful in evaluating general performance, unlike the mostly technical and specialised benchmark suites one more usually encounters.
0 0 [Posted by: MTX  | Date: 02/04/09 01:43:26 PM]
Reply
 
From Phoronix site:
Runs On Linux, OpenSolaris, Mac OS X, & FreeBSD Operating Systems
Too bad
0 0 [Posted by: Gavric  | Date: 02/04/09 02:26:13 PM]
Reply
4. 
At the end of the conclusion of your report, you stated " it almost catches up with Core 2 Quad Q6600 here " What motherboards were you using that you were to get 3.72 GHZ. I would be interested in knowing which motherboards you recommend inasmuch as my Intel DP35DP is so so.

Thanks
0 0 [Posted by: ambertape  | Date: 02/03/09 07:17:35 PM]
Reply
- collapse thread

 
I have used ASUS P5Q Pro for LGA775 and Gigabyte MA790GP-DS4H for Socket AM2+. I could recommend any boards from Gigabyte EP45 series also. We have tested most of them, they overclocked great.
0 0 [Posted by: Gavric  | Date: 02/04/09 04:47:49 AM]
Reply
5. 
AMD is going down, and is going down fast. I hope they wont drag ATI along...
0 0 [Posted by: TAViX  | Date: 02/03/09 10:39:08 PM]
Reply
6. 
Why is this article titled "New Overclocking Star"? I mean you took this thing to....ummm..3.7Ghz? Not exactly impressive. Shouldn't it be called "New Overclocking Star?" because obviously it's nothing special.
0 0 [Posted by: Pixelated  | Date: 02/04/09 05:34:05 PM]
Reply
7. 
They should have ran the games at 800x600 and low quality. It looks as if the graphics card is the bottleneck, when the reviewer should take steps to ensure that the CPU bottlenecks in an article testing CPUs.
0 0 [Posted by: hedron  | Date: 02/05/09 08:39:58 AM]
Reply
- collapse thread

 
u idiot ! who will buy an expensive processor and run games on it on a resolution of 800 x 600?
0 0 [Posted by: PFX  | Date: 02/11/09 10:05:48 AM]
Reply
8. 
So much for being a star. I actually though for a minute that AMD had a chance when I came saw this article. 3.7, really? why don't you try an extreme form of cooling on the side to see where it's limits really are. If it can't go past 4 on water it is a waste of time trying to get any performance out of it. Also, why didn't you overclock the I7? I got mine to 3.7 in about 10 minutes after installing it. by 30 minutes it was at 4.4 (on water) at 65c, 75-80c load.right now mine is set to 3.75 running at 40c, 50c load. try using a better stess test that isn't made for an intel cpu in the first place, or the amd will naturally be way behind because of low cashe mem.
0 0 [Posted by: Shckr57  | Date: 02/11/09 03:51:27 PM]
Reply

[1-8]

Back to the Article

Add your Comment


Our Sponsors

Crucial.com Western Digital G.Skill