Neoseeker : : : : AMD Phenom 9900 Review

AMD Phenom 9900 Review - PAGE 2

- Wednesday, December 19th, 2007 Like Share






Get updates when we publish new articles

Comments

Sort by date: ascending descending
0 thumbs!
^
skatcat31 Dec 29, 07
Well right now it's got the B2 bug, but wait for B3, and it will destroy intell. Again.
0 thumbs!
^
Hiko Seijuro Dec 25, 07
is the 9900 realy as it was ment to be.. or are they pullin old intell spiting things out before finalizing it...
0 thumbs!
^
skatcat31 Dec 24, 07
It doesn't state that they used the patch or not, it jsut mentions it. Also the TLB patch is a great way to elp keep your stepping stable, een a the performance hit.

Two things to keep in mind:
1. Wait for hte B3 stepping. Seirously
2. The intell won because it's a higher standard chip, meant to be for gamers. The Phenom was meant for buisinessed, althgouh stil nice.
3. Wait for hte Pheonix, or whatever the gaming desktop edition will be called now.
0 thumbs!
^
enumae Dec 22, 07
Could you tell me if this review was done with the TLB patch applied?

If I had missed where you clarified this my apologies.

Thanks
0 thumbs!
^
skatcat31 Dec 20, 07
I've gotta say, this is a nicer test then the one for the 9600. Unfortinatly from the reveiw of the MSI board, I knew there were problems there, and my ugly tired reared head showed itself. however for these results I'm acutally muc mroe satisfied. Now only to wait for B3 stepping ot get rid of the problems and yet again, AMD will sit kindly ontop. they may take long, but having native quad core, it kinda makes intell look silly. Also remember, windows does do auto stepping during performance riddled tasks. I've found windows overclocking my CPU before from 2.0 Ghz x2, to 2.56 Ghz x2, something not even my bios could do with my RAM chip...(which makes me wonder). Also it's nice to finely see that AMD is yet again putting ocnsideration into their chips instead of just putting mroe di onto single chips. YAY FOR HEAT ENVELOPES! For hte fact that it obtained 3.0Ghz with the stock cooler, im mightly impressed and can only wait for hte next generation stepping to build my next rig. DDR3 AMD buisiness benchmarks anyone?

edit: also cisco kid, the xenon was more for price range refence and if you compare the stock timings, it's a good reference at that. In the same price range, nothing kept up with the AMD phenom. Overclockign it, well that's something more users do on their own. In fact it was even stated they included it purely for the fact that it was in the same price range. Of course an unlocked 1000 USD chip is gonna trounce something that's nto even half it's price. Jsut wait untill AMD releases the Phenom FX. didn't they tought the new FX name ot be Phenox? or Phenoix or something... I could never remember my mythical creatures... and to think I saw a unicorn the other day.
0 thumbs!
^
bhenning Dec 20, 07
I compared to the quad core results I had from previous results; and the comparison was based on stock speeds.

ofcourse an overclocked X3210 will beat an overclocked Phenom 9900 - read my review of the X3210 if you want to see the stable OC I achieved - but that was not the intent of the review, the intent was to compare stock performance, and see how high the 9900 would clock up.

I included the higher priced QX6700 as it runs at almost exactly the same stock speed, and I was frankly surprised as to how well the 9900 held up against that much more expensive chip.

quote cisco kid
The review is klinda lame, we are talking about a $350 cpu against $247 xeon x3210. Of course at stock speeds one can not expect the xeon 3210 to compete with a core speed of 2.13 vs 2.6 come on. However the 3210 would smoke the Phenom had you been fair and overclocked it, they are easily capable of running 3.5-3.6 if a G0 stepping. Mine is running 3.4 rock solid on default vcore with 4,4,4,12 1T using GSkill HZ PC6400 on a 680i evga sli board, with one bump in vcore I have 3.5 solid

I would love to see results with all chips overclocked cause Phenom would be clearly smashed by a 3210 at $100 less

0 thumbs!
^
bhenning Dec 20, 07
Thank you for the correction!

I looked at the wrong bar on the chart. Mistakes happen... I'm correcting the article.

Best,

Bill

quote dricks
quote
TMPGEnc

The Phenom 9900 beats both the Core 2 Quad QX6700 and the Xeon X3210 at stock speeds here!
Since when does 284s (Phenom@stock) is better than 268s (QX6700) and 282s (X3210) when comparing encoding time?

0 thumbs!
^
cisco kid Dec 20, 07
The review is klinda lame, we are talking about a $350 cpu against $247 xeon x3210. Of course at stock speeds one can not expect the xeon 3210 to compete with a core speed of 2.13 vs 2.6 come on. However the 3210 would smoke the Phenom had you been fair and overclocked it, they are easily capable of running 3.5-3.6 if a G0 stepping. Mine is running 3.4 rock solid on default vcore with 4,4,4,12 1T using GSkill HZ PC6400 on a 680i evga sli board, with one bump in vcore I have 3.5 solid

I would love to see results with all chips overclocked cause Phenom would be clearly smashed by a 3210 at $100 less
0 thumbs!
^
dricks Dec 19, 07
quote
TMPGEnc

The Phenom 9900 beats both the Core 2 Quad QX6700 and the Xeon X3210 at stock speeds here!
Since when does 284s (Phenom@stock) is better than 268s (QX6700) and 282s (X3210) when comparing encoding time?
0 thumbs!
^
bhenning Dec 19, 07
The text by the charts, and in the conclusion, was comparing stock speed results. The overclocked results were also shown, and clearly identified with "200x15" on the bar chart labels.

Thanks for the heads-up. I fixed the few minor labeling errors.

quote BFELLOW
You need to make it clear that one of the 9900's is overclocked to 3.00ghz and all the inaccurate statements about the Q6700s losing to the 9900 at stock speed in a few of these graphs. You also confuse the 9600, 9900, and Q6700 in some of your statements.


This message was edited by bhenning on Dec 19 2007.
0 thumbs!
^
bhenning Dec 19, 07
Hi Jerry,

I realize the B2 stepping is not new - however I believe the 9600 I tested earlier was a B1 stepping; and I agree, I was lucky to get a B2 that was stable at 3.0GHz

When I did a check on NewEgg before publishing the article it showed a $949 price tag; I just repeated the search, and that was the first price for the boxed extreme processor. http://www.newegg.com/Product/Product.aspx?Item=N82E16819115011
The price you pointed at is the non-extreme SKU, I looked up the SKU for the one I tested in the earlier review. I do appreciate the heads up and will update the article to show that the non-extreme edition is available for $560.

No need to be perplexed, I simply summarized all the tests I ran, and I am not willing to ignore any of them. We've been using the same benchmark suite for quite a while now (with occasional minor changes, like recently dropping non multi-threaded WinRAR, POV and Lame); I did not want to restrict the scoring to gaming only. I agree that the much larger cache on the Intel helps it.

I wish I had time to run more benchmarks, and compare different Intel chips - however I had a total of three days to do the review, and had to use Intel results that were already in our system from past reviews.

Best Regards,

Bill

quote Jerry_A
Overall interesting review.... unfortunately, there are a hand full of inaccuracies...

First, you assume the 3.0 GHz OC came from a new stepping, but the B2 stepping was the launch stepping -- check other reviews it is certainly B2 -- B3 will be out mid-late Q1 07, and per AMD the higher clocked parts (this part) will not be out until the April-May timeframe (see transcript of analysts day last week). You may have been lucky, one or two sites were able to get 3.0 GHz stable enough to bench, but not meany... generally topping out at 2.8 GHz or so.

Next, the QX6700 is an unlocked version of Intel's 2.67 GHz quad, the retail version is actually pricing in at $560 or so http://www.newegg.com/Product/Product.aspx?Item=N82E16819115027 still not cheap but running at stock, this is what could be had, a 2.5 Ghz yorksfield will launch next quarter which will come in at a rumored 250 bucks, and all indications is that will be roughly the same as a QX6700.

I am perplexed by the fine binned detail you use for summarizing 1st places, if you ignore the synthetics, right mark, Sandra, and winstone -- the score is not quite in the 9900's favor: 3 W for 9900 and 9 for QX6700 ... it is a bit misleading to show the data they way you did when it is clear mem BW will be favored on the IMC, but on DT the IMC is not a huge factor compared to a larger cache pool on the Intel chip.

It is too bad you did not take a multiplier unlocked 9650 and clock it down to 2.5 GHz (the 250 dollar Yorky intro next month or so), that would have been as informative.

Finally, your results are not that far off with other reviews, but there are higher quality reviews on the net:
http://www.bit-tech.net/hardware/2007/12/13/amd_phenom_9500_9600_9700_and_9900/1 (they are showing a 9900 more or less even or losing to a Q6600)





This message was edited by bhenning on Dec 19 2007.
0 thumbs!
^
BFELLOW Dec 19, 07
quote bhenning
Yes guys, another Phenom review!

And this time... the chip is a lot more competitive.

http://www.neoseeker.com/Articles/Hardware/Reviews/phenom_9900/
You need to make it clear that one of the 9900's is overclocked to 3.00ghz and all the inaccurate statements about the Q6700s losing to the 9900 at stock speed in a few of these graphs. You also confuse the 9600, 9900, and Q6700 in some of your statements.
0 thumbs!
^
Jerry_A Dec 19, 07
quote guticb
My only complaint is that a Q6600 wasn't benched and compared to it. I want to see how it compares to one of those beasts.
Here are a couple for you:
http://www.legitreviews.com/article/597/9/
http://www.hexus.net/content/item.php?item=10648
0 thumbs!
^
Jerry_A Dec 19, 07
Overall interesting review.... unfortunately, there are a hand full of inaccuracies...

First, you assume the 3.0 GHz OC came from a new stepping, but the B2 stepping was the launch stepping -- check other reviews it is certainly B2 -- B3 will be out mid-late Q1 07, and per AMD the higher clocked parts (this part) will not be out until the April-May timeframe (see transcript of analysts day last week). You may have been lucky, one or two sites were able to get 3.0 GHz stable enough to bench, but not meany... generally topping out at 2.8 GHz or so.

Next, the QX6700 is an unlocked version of Intel's 2.67 GHz quad, the retail version is actually pricing in at $560 or so http://www.newegg.com/Product/Product.aspx?Item=N82E16819115027 still not cheap but running at stock, this is what could be had, a 2.5 Ghz yorksfield will launch next quarter which will come in at a rumored 250 bucks, and all indications is that will be roughly the same as a QX6700.

I am perplexed by the fine binned detail you use for summarizing 1st places, if you ignore the synthetics, right mark, Sandra, and winstone -- the score is not quite in the 9900's favor: 3 W for 9900 and 9 for QX6700 ... it is a bit misleading to show the data they way you did when it is clear mem BW will be favored on the IMC, but on DT the IMC is not a huge factor compared to a larger cache pool on the Intel chip.

It is too bad you did not take a multiplier unlocked 9650 and clock it down to 2.5 GHz (the 250 dollar Yorky intro next month or so), that would have been as informative.

Finally, your results are not that far off with other reviews, but there are higher quality reviews on the net:
http://www.bit-tech.net/hardware/2007/12/13/amd_phenom_9500_9600_9700_and_9900/1 (they are showing a 9900 more or less even or losing to a Q6600)


^
Sponsored
Sort by date: ascending descending
Add your comment:
Name *:  Members, please LOGIN
Email:  We use this to display your Gravatar.

Sign in with
Comment *:
(0.3355/d/web8)