AMD Phenom 9600 Black Edition - A New Hope?

Status
Not open for further replies.

enewmen

Distinguished
Mar 6, 2005
2,247
3
19,815
Too bad. The Phenom came after the Kentsfield and should perform better.
Lets hope this is a manufacturing problem and not an architectural problem.
 


It could also be a CPU/BIOS/board tuning problem as well. The scaling numbers when overclocked weren't very good and suggest that the reduction in memory speed causes a big hit in any additional performance that overclocking may yield. The processor is supposed to be able to run RAM at speeds up to DDR2-1066 but for some reason it wouldn't do over DDR2-800 with their setup. I'd like to see what the system would do with RAM running at ~DDR2-1066 rather than ~DDR2-800.
 

SirCrono

Distinguished
Sep 9, 2006
463
3
18,785
Yay, a sub-par CPU with an unlocked multiplier... too bad it stills OC worse than the Q6600.

I really hope AMD gets their act together soon.

 

coret

Distinguished
May 29, 2007
273
0
18,780
This is the third 9600 BE thread, the first being the one i posted December 20th ... the second being thunderman a week or so ago ...

muk, you're either thunderman's new handle, or just really like beating dead horses.
 

spongebob

Distinguished
Apr 23, 2004
335
0
18,790
Or perhaps a Tom's employee who just started a thread about a THG article? :??:
 

black9ice

Distinguished
Jan 24, 2008
1
0
18,510
1) The thread was started in referance to a Tom's article.

2) It still amazes me through all of the bashing on the latest AMD CPU parts that not one, and I have checked!, not one site reminds us these are first gen parts for Phenom, nobody seems to test with 1066 ram, and all fail to mention the fact that all current Phenoms use HT 3.0 but only at 3600 and not 5200!!

3) I am not a fanboy, gayboy, greyboy. I am not paid by either realm. I want faster, cheaper, less power hungry parts. If you clearly look at AMDs road map I think that is where they are headed.

4) I put on an extra 200-300 kilowatt hrs a month this past summer because of my stupid "quad-corez Roxors your pants" Intel/Nvidia machine. It is all sold and I am now using a AMD BE 5000+ and 3850 256meg to game with at 16X10 and it performs just fine!! JUST FINE!!
 

Slobogob

Distinguished
Aug 10, 2006
1,431
0
19,280
I'm shocked. How can a company be so stupid?

I speculated about it in december when i made my thread about the release of the 9600 BE and sadly AMD made every mistake they could. (http://www.tomshardware.co.uk/forum/246610-10-phenom-black-edition-announced-2007)

1. They used the B2 stepping. That's borderline insane.
2. They priced it way to high.

I've read this article and the conclusion was quite familiar. AMDs CPU products are no longer competitive at their current price. That is quite alarming.
(http://www.anandtech.com/cpuchipsets/showdoc.aspx?i=3153&p=11)
(http://www.xbitlabs.com/articles/cpu/display/dualcore-shootout_11.html#sect0)

A good way to see the real value of something is ebay. I've monitored Phenom prices there and if the bidding starts at 1€ it usually stops at around 140-150€ - no matter if it's the 9500 or 9600. This indicates that people are not paying for the performance of the processor but for the product name or better "they buy it just to have it". It's like someone bidding 450€ for a Pentium D 840 EE. Most core 2 duos are faster at stock than those PD EEs but they get bought because they are unique, like collectors items. The same is true for the Phenom.
The gap between 2 and 4 cores and the performance difference between the two is seriously messed up. A x2 (windsor or brisbane core) CPU costs me roughly 50€ while a Phenom CPU costs close to 160€. That's quite a steep increase considering a performance increase of about 5%-10% comparing core to core.
Intels Q6600 costs about 190€ while a 6550 or 6750 cost only 130€ or 150€ (The E6600 is no longer sold by Intel!). They perform on par core to core but the markup is not nearly as high as that seen with AMD.

I know the platform of AMD is a little cheaper. A feature rich am2+ board is cheaper than most comparable socket 775 boards, but that shouldn't and can't outweight the odd processor prices.
The power consumption is another issue and another field AMD has lost to the competition, that needs to be taken into account regarding the price. (See links)

What should a Phenom cost? What is it really worth? 120€ (175$)? 140€ (200$)? Or more like 100€ (145$)?


 

Crashman

Polypheme
Former Staff
AMD keeps talking about the upcoming fixed cores, yet they still haven't delivered any. Still, they send out review samples of the current core, hoping that "any" publicity will help. Reviewers are obligated to review the samples they receive, yet are also obligated tell the truth about their findings.

Everyone on the consumer side wants AMD to succeed, but it appears they simply aren't up to the task at this time.
 

Hatman

Distinguished
Aug 8, 2004
2,024
0
19,780
I want too see the 45nm ones. The only thing that would make me consider AMD for my nexdt build would be those with much higher clocks.

I generally prefer AMD to Intel but nto if they put out products like these. For overclockers they are just pathetic.
 

sailer

Splendid


Though I can't say anything about payment, this article certainly did come across pro Intel. Instead of simply telling about the Phenom, it was constantly bringing up the idea, "the Phenom 9600 BE isn't as good as an Intel Q6600". I think everyone knows that already. All impartiality was lost very quickly, and instead of telling people what could be accomplished by the 9600 BE in comparison to a standard 9600, it was that consistent, "it isn't as good as Intel's Q6600" that destroyed the validity of the article. It also seemed that the overclocking was done purely by the multiplier, with no attempt made to increase the FSB speed. That leaves me wondering if any higher speeds could have been achieved through a combination of a higher multiplier and a higher FSB. I don't know that it could, but I do wonder.

And yes, I would like to see a review that used a top end mobo and ram that ran at the rated DDR2 1066 speed instead of using ram that can't even be brought up to DDR2 800. While that won't help the Phenom itself, it could mean higher scores in the benchmarks.
 

caveira2099

Distinguished
Jan 11, 2008
14
0
18,510
WOW, yet another USELESS THG article!!!

I am tired of THG Intel-favored opinions. I have written many times here at the forum and to the site staff that one cannot simply toss some (WRONG) results to justify an OPINION.

THG again shows mistakes in their tests, proving their COMPLETE LACK of knowledge of what they are doing. They insist on using the wrong version of Sandra to benchmark this new AMD arch, just as an example.

I can't even be tagged as a fan of any of those brands - my computers aren't even x86 anyway... I like reading CPU news, but only when they are at least trustable. I thought THG was, but I was wrong...

No more THG for me. As for you BLIND people who still believe in THG's articles, my regrets.

P.S. Just checked intel.com to see a self-add banner citing THG... Maybe Intel money is driving the authors here....
 

ragemonkey

Distinguished
Jun 26, 2006
186
0
18,680
There are several overclocking articles written on the 9600BE. This one is by far the worst that I read. I agree with the above critique about no 1066 memory and no reference clock adjustments. Some of the other articles do both. Some have had more success raising the ref. clockm while some had more with the multiplier alone. Most of the articles say the same thing. The 9600BE is a complete gamble. You may get one that OC's to 3.2... or you may get one that won't even OC to 2.4.

I bought one of these to test out. Mine doesn't even run stable at the stock 2.3ghz speed. Every so often I get a BSOD about a secondary processor not responding in the allocated time. I figure the chip I got has at least one flaky core. I've RMA'd it to newegg and hopefully my next one will have better results.

When it does work it works very well... and the difference in overall system response is much better than my 6400BE. Benchmarks only say and show so much. I prefer the actual "look and feel" test as opposed to raw numbers.
 

wolverinero79

Distinguished
Jul 11, 2001
1,127
0
19,280
Is THG reporting news in an Intel favored way, distorting reality, or is reality that Intel is doing awesome compared to AMD and THG is just reporting what reality is? I tend to think it's more towards the latter.
 


I agree 100% except the statement about being cheaper. For the same price as a AM2+ mobo you can get a P35 that is compatable. At least one that is decent with nice features such as dual PCI Express and 6+ SATAII ports. I don't do cheap unless its for those who only use internet. For enthusiasts most AM2+ mobos are around the same price as a P35/X38 mobo.



Wow. Thing is that most sites have even shown that when OCed to 2.6GHz (Phenom 9900) it is not able to beat a Q6600. No the difference was only 0.3% overall but that goes to show that the Phenoms IPC is not as great a C2Q. Basically in order to get the same performance in an application with a Phenom as a Q6600 you would need a 2.7GHz Phenom which sounds about right.

THG is not Intel biased. Hell when I came here when AMDs X2's were on top all the articles would talk about how Intels offering didn't beat AMDs. There were a few now and then back then where Intel would keep up but not beat them. Its an ENTHUSIAST site. Whatever performs will be whats the most prefered. Hell right now for the price a 3870X2 beats in some and keeps up with a 8800 Ultra but form about $300 less giving it the prefered GPU. But before that the 8800 Ultra was the top and prefered.

BTW why don't you post us a link showing that banner? I didn't see any when I went to Intels site. Just their own stuff.

And just to compare, I have my Q6600 OC'ed to 3GHz(25% OC) and the CPU Voltage is at 1.248v. I like how they show that not every Phenom BE will OC to a specific speed, unless you are really lucky, where as pretty much every Q6600 is able to. Too bad they didn't OC it as well to the same speeds to test the IPC's for each CPU. The truth hurts. Phenom isn't competative really. Not a bad upgrade from a older X2 but still not the performance you would expect.
 

SEALBoy

Distinguished
Aug 17, 2006
1,303
0
19,290
This result was entirely expected... B3 needs to come and fast and be able to hit ATLEAST 3.0GHz. I think it's sad that AMD's new tech is still slower clock-for-clock than Intel's 18-month old technology...
 

keithlm

Distinguished
Dec 26, 2007
735
0
18,990


Since the forum post was online pointing to the link a few minutes before the link was actually available on the site... I'd say you are correct. (I was online in the forums just at that moment and had just hit refresh. Saw the new post... but the link didn't go anywhere for a few minutes.)
 

Mathos

Distinguished
Jun 17, 2007
584
0
18,980
Good god, even I have better results out of my 9600BE. Running 2.6core 2.4nb at 1.248v according to Everest still with no problem. And the sad part is some of the benchies he's running would be affected by the extra bandwidth from boosting the memory controller, though it wouldn't help as much with the game benchies, well guess that would depend how memory intensive some games were though..

Oh yeah, and as long as the nb/IMC is left at 1800Mhz t the base memory speed will be 750Mhz or around there. My memory has been running at 800Mhz since I put the NB up to 2.0ghz, it may be due to what setting he uses for memory speed in the bios though. I know with the k9a2 platinum you either need to set it to 1:2 for ddr2 800 or auto works as well. For 1066 you need to set it manually to the next higher setting and up the ram voltage to the specs for the ram. IF he's running stable at 2.6ghz on the gigabyte board, I'd have to suggest using the following settings in the P-states menu on the k9a2 platinum.

CPU FID=10, VID=23, DID=divide by 1
NB FID=8, VID=24, DID=Divide by 1

Those are the exact p-state settings that I am currently running, on bios 1.1b3 I was doing 2.7ghz on bios 1.2 but with the performance hit of the TLB fix.

I've had no luck ocing with anything set to divide by 2 or higher, this is also how the C&Q is able to use .25 multies.

It also helps as to whether C&Q is disabled in the bios settings, as well as disabling spread spectrum. Both of those will have an impact on OC stability with this processor, especially with a crappy psu like my raidmax.
 
Status
Not open for further replies.