Sign in with
Sign up | Sign in
Your question
Closed

Six-Core Analysis: AMD’s Phenom II X6 Gets Scaled Down

Last response: in Reviews comments
Share
June 24, 2010 6:43:06 AM

thats a nice motherboard
Score
7
June 24, 2010 6:56:30 AM

WinZIP should just be dropped from the benchmarking suite until a multithreaded version comes out. It's always the same, flatline graph.
Score
25
Related resources
June 24, 2010 7:04:44 AM

Interesting article, thanks for the work.
Score
9
June 24, 2010 7:11:35 AM

randomizerWinZIP should just be dropped from the benchmarking suite until a multithreaded version comes out. It's always the same, flatline graph.


Speaking of graphs.

Toms - The majority of your graphs are in descending order (6 cores to 1), but some of them are ascending. Was a little confusing for a moment until I realized the switch up.
Score
20
June 24, 2010 8:09:14 AM

good job on the tests! i have the same complaint: using a 750w psu with a system that uses between 80 to 200watts. a 300-400 watt psu would be best.

the system is idling at 80watts and the system is using a velociraptor (which is not that efficient) and a discrete gpu (idling at ~15watts)... this means you can get this baby to idle at 60watts or so if you use the onboard gpu and a more efficient hdd... and if you use a proper size, efficent psu then this could get even lower. that is so sexy! one could make a silent and efficient system with this. no wonder thubans are in short supply.
Score
6
June 24, 2010 8:11:46 AM

and yes... drop that stupid winzip from tests. hardly anybody using that. you can make zips with windows xp. if you are lazy you just trial winrar (indefinetly :) ) and if you are not lazy you get 7zip.
Score
9
June 24, 2010 9:04:34 AM

Winzip is dead. Can you use 7zip from now on please Tom!
Score
15
Anonymous
June 24, 2010 9:07:35 AM

Just replace winzip with 7zip it support multi thread if i recall
Score
10
June 24, 2010 9:32:19 AM

would have liked to see how much further you could overclock this thing if you're only running 3 or 4 cores.
Score
9
June 24, 2010 9:37:39 AM

Thing with lame is that nothing limits you to run it only once, I am usually starting at least 4 of them at once, each using one of my 4 CPU cores. This way it scales even better then most multithread applications.
Score
1
June 24, 2010 9:53:34 AM

freaking GRAPH INCONSISTENCY !!!! this seems to be the curse of this site. No one of the reviewers thinks about proper data presentation. If you are graphing 6 to 1 cores performance numbers, do it consistently. that way I can just look at the best result in the graph nad see which one was worst and which one was best. now I need to read the column labels anyway in each graph just to make sure I get the correct information.
Score
6
June 24, 2010 9:56:04 AM

also what's with the Cinebench single threaded graph ? all have the same result except single core but top to bottom it goes 4 3 6 5 2 1 !!! WTF !!!!
Score
1
June 24, 2010 10:00:35 AM

Normalized Power And Efficiency Results page is all nuts ... the unit is % not Watts

realy crap data presentation in the whole article ...
Score
2
June 24, 2010 10:08:23 AM

haplo602also what's with the Cinebench single threaded graph ? all have the same result except single core but top to bottom it goes 4 3 6 5 2 1 !!! WTF !!!!

It's ordered from best to worst (top to bottom), but as the graphs are probably auto-generated the labels aren't taken into account and equivalent results don't have any particular order.
Score
2
June 24, 2010 11:04:40 AM

Interesting article.

A big jump from a single core to a dual ...

Well done.
Score
1
June 24, 2010 11:13:00 AM

Off-topic but I would love an educated answer..
Any news on whether 'Fusion' PPU's will work on existing MB's?
Also I have a doubt about Liano or whatever (the desktop part).. They say that it has the processing power of an HD5770 for graphics calculations but at the same time it's only one unit doing all the maths.. so does this mean that it can reach that goal only when it doesn't have a lot of CPU jobs?
Is it meant to be the winning competitor of SandyBridge or is that what Bulldozer is all about?
Thanks in advance..
And I'd also like some sources for the answers you provide..
Score
-7
June 24, 2010 12:03:00 PM

Hmmm, it may not be the best "gamer-CPU," but for those less interested in maximum overclocks (such as myself), and/or who multi-task a lot or use well-threaded apps, this chip looks very nice.
Score
2
June 24, 2010 12:42:22 PM

Any 'true' reason for the thump down?
Score
-5
June 24, 2010 12:51:04 PM

Nice article but nothing really shown that 95% of Toms readers didnt already know fairly well.

As Toms has stated a thousand times if you need more cores buy the model with them if you dont stick with the lower core model and save your money, anything else is a waste of money or a crapshot.
Score
1
June 24, 2010 1:12:54 PM

so, 4 cores saturate the memory controller. Then it would be interesting to test that with different memory clocks.
Score
5
June 24, 2010 1:46:30 PM

Nice article!!!
I like to see the core i7 980x with the same test, with HT enabled and disabled for each core (1, 1HT, 2, 2HT, 3, .... ,6, 6HT).
A bit laborious, but I find it very interesting to watch.
Score
2
June 24, 2010 1:48:56 PM

Another vote for 7zip to replace Winzip please.

psycho sykesAny 'true' reason for the thump down?


As you stated...it's off-topic.
Score
8
June 24, 2010 1:50:59 PM

GFGNice article!!!I like to see the core i7 980x with the same test, with HT enabled and disabled for each core (1, 1HT, 2, 2HT, 3, .... ,6, 6HT).A bit laborious, but I find it very interesting to watch.


I'm curious about hyperthreading in modern-day apps as well. Maybe I missed something, but hasn't it been ages since you guys have done a side-by-side comparison about hyperthreading on versus off?
Score
4
June 24, 2010 1:56:36 PM

So in other words, set this thing up for 6 cores active for multi-threaded apps, 3 cores for gaming, and 1 core at idle.

I just ordered mine!
Score
1
June 24, 2010 2:17:35 PM

No, you can leave all the cores on when idle, and the power consumption is not notably higher.
Score
1
June 24, 2010 2:28:29 PM

psycho sykesAny 'true' reason for the thump down?

Because fusion will be a notebook part? And it is code named Llano.
The desktop part is Zambezi and it does not include fusion.
And how do you expect us to give you sources if you don't give us sources for what you heard?
But that's no real reason for a thumbs down. The Thumbs down is because you asked your questions here instead of in the forum where people are there to help you.
And I didn't give you a thumbs down :) 
Score
3
June 24, 2010 2:32:59 PM

Can I please get an article comparing a 1055t stock and overclocked to a 1090t stock vs overclocked? I want to buy a 1055t if I can get it close to a 1090t performance. You do articles like it all the time. I just read an article pushing a six core intel chip to 5ghz. Which is cool, but hardly practical.
Score
1
June 24, 2010 2:54:57 PM

Very good article, the 1090T will be the replacement of my X4 955 that is installed in my Gigabyte 890FXA-UD5. I just want wait a little more for the 1090T C3 or something like that.
Score
0
June 24, 2010 3:01:07 PM

Here's an interesting benchmark that I never see.

Show me peak power performance on individual tasks.

Yeah, using 100% of 6 cores has a lot of power usage compared to one, but what does the power usage look like when I run winzip with all 6 cores enabled?

I would assume that peak power usage would be equal to 100% of 1 core usage but there's nothing in the benchmarks or text that says so. If most of my tasks are single threaded, do I get a power benefit from turning off cores when operating those single threads at 100%?
Score
1
June 24, 2010 3:09:24 PM

psycho sykesAny 'true' reason for the thump down?

Because the amd fandois don't need a reason. I got thumbed down cause I said we need more nvidia chipsets for amd. I promise you that if nvidia made a 1156 chipset no one would want p55
Score
-9
June 24, 2010 3:10:55 PM

Very well-written article; thank you very much.
Score
1
June 24, 2010 4:33:50 PM

Awesome article, I really enjoyed it, thanks!
Score
1
June 24, 2010 5:01:58 PM

to psycho sykes, the reason you get thumbs down is because this is for comments on the article, not a forum for you to lazily get information. Google, search the forums on THG, and get your answer, don't clog article comments. Now you made me do it...

Good article though, I too agree that the data should be shared in a consistent format, but it's not as bad as people are making it out to be, just takes a minute to read what bar goes to what item.
Score
1
June 24, 2010 5:46:00 PM

where are the gaming tests?

all the tests that were performed (with a few exceptions) were things that would obviously scale well with multiple cores.. so its not really a big surprise that efficiency went down

when i turn on my pc i dont do it to convert video, run 3d mark or encrypt something.. i use it to do normal everyday things like check my email, play videogames, browse the web, watch movies, etc. and when i on occasion do convert/encrypt something i dont care if it takes 1 min or 2 min... its still the time it takes for me to get something to eat or go to the bathroom or hell, play spider solitaire.

what im trying to say is i wish there were more tests that reflected things we do everyday. i know these would be hard to make scientific but here are some suggestions:

-browser load times
-other program load times
-boot time

these tests better reflect how good your pc experience is.

how does multiple core scaling affect that?
Score
4
June 24, 2010 5:54:40 PM

Tx: Patrick Schmid and Achim Roos

Nice Article :) 
Score
0
June 24, 2010 7:46:24 PM

No games testing???
Score
3
June 24, 2010 9:00:04 PM

To all who replied to me..
@bourgeoisdude: I said it's off-topic and asked gently what I wanted without really making a thump-down worth mess here..
@Enzo Matrix: The source is an article here on Tom's (or maybe a link to it) that I can't find but I am pretty sure it was either a news article or a link in a news article..
I'm 100% sure that there was talk about a desktop counterpart of LIano.. Can't find the source again though (it's the same previous one)..
@jomofro39: I just can't believe that you thought I didn't use Google before asking here.. and after all a more educated answer in-the-know will get me what I need to know more precisely.. most TH's users know that Google is their best friend..
@Poisoner: Sorry, but I wasn't really fanboy'ing.. I think however that nVidia chipsets are more suited for Intel's almost non-existent built-in GPUs.. AMD is doing pretty good on the other hand.. and honestly, I would never buy an nVidia MB for a discrete GPU..
--------
And I didn't really feel like making a thread for one simple question.. the question came to me and I posted it.. I only wanted to keep forums clean (some friends' advice) so I am sorry for that..
Score
-2
June 24, 2010 9:50:45 PM

i think the hexa core has been rather good and the specs are good
Score
0
June 24, 2010 10:06:17 PM

I love the MP2 to MP4 compression on the 1090T it's the King of the hill for the price.
Score
0
June 24, 2010 10:08:26 PM

saint19Very good article, the 1090T will be the replacement of my X4 955 that is installed in my Gigabyte 890FXA-UD5. I just want wait a little more for the 1090T C3 or something like that.


That's a good mobo, sound similar to what I did but with the 890FXA-UD7. Either way it's a nice upgrade to the 1090T.
Score
0
Anonymous
June 25, 2010 12:13:21 AM

Sure, some apps aren't optimized for many cores... but how about a performance test that takes several single-threaded apps and runs them at the same time? If they take 10 minutes to finish serially, and you run them at the same time on a multi-core, how much performance advantage? That's more real-world.
Score
2
June 25, 2010 1:19:11 AM

"Since Apple and Adobe aren’t exactly nobodies, this fact disappoints even more."

Sorry, I can't help but snipe at that one. I keep as far away from products of both companies as I can, and my PC thanks me for it vigorously. They are exactly nobodies in my book.
Score
3
June 25, 2010 3:15:32 AM

good test for the 6 core, showing its use. this is great value for people using professional and video apps.
in my case, can't saturate my AMD dual core processor for online use and occasional gaming.
Score
2
June 25, 2010 3:31:46 AM


If one ran these tests in the same way on a 980X, would
one see the same differences in performance for each
application? ie. does each app in any way favour a
particular CPU arch? I suppose this also begs an answer
to the HT vs. non-HT question.

Ian.

Score
0
June 25, 2010 4:27:13 AM

psycho sykesTo all who replied to me..@bourgeoisdude: I said it's off-topic and asked gently what I wanted without really making a thump-down worth mess here..@Enzo Matrix: The source is an article here on Tom's (or maybe a link to it) that I can't find but I am pretty sure it was either a news article or a link in a news article..I'm 100% sure that there was talk about a desktop counterpart of LIano.. Can't find the source again though (it's the same previous one)..@jomofro39: I just can't believe that you thought I didn't use Google before asking here.. and after all a more educated answer in-the-know will get me what I need to know more precisely.. most TH's users know that Google is their best friend..@Poisoner: Sorry, but I wasn't really fanboy'ing.. I think however that nVidia chipsets are more suited for Intel's almost non-existent built-in GPUs.. AMD is doing pretty good on the other hand.. and honestly, I would never buy an nVidia MB for a discrete GPU..--------And I didn't really feel like making a thread for one simple question.. the question came to me and I posted it.. I only wanted to keep forums clean (some friends' advice) so I am sorry for that..



I wasn't saying you were being a fanboy. I was commenting on how the AMD fanboys around here love to thumb down everything they can. Worst part is that I have a kick ass AMD budget box that I just added an HD 4870 to.
Score
-3
Anonymous
June 25, 2010 8:26:30 AM

In real world, we have multiple processes/applications running all the time when using a PC - messenger, anti-virus, browser, music player, etc.

How does multiple core CPU handle multiple processes/applications running simultaneously that only runs single-threaded like Winzip? Are those processes confined to the first core or distributed across the different cores? If distributed, then there's really no benefit on benchmarking single-threaded apps since in real world those apps would be running simultaneous with other processes anyway ... and this means that in real world computing, you'll always get better performance from more cores, since nobody opens a PC and just use one application at a time.
Score
1
June 25, 2010 8:35:28 AM

Core for core Intel is 2x faster.

55 GFLOPS? Intels 6x Gets 109=GFLOPS.

Score
-1
June 25, 2010 8:49:14 AM

raptormnl, it partly depends on how smart the OS scheduler is at
assigning threads to cores, and especially on ensuring threads do
not bounce around between cores unnecessarily. This generally
works better on UNIX variants, especially the highly scalable
versions such as IRIX, though I think mainstream Linux (Red Hat,
Suze, etc.) as absorbed a lot of this technology now (eg. the
1024-CPU i7 XEON Altix UV). I've also read that Windows7 is
better than XP/Vista at managing threads, but I've not seen any
proper reviews to show this is true.

Is it possible in Windows to force a particular application to
run on a specific core? Under IRIX I just use the 'runon' command. Do ordinary Linux variants like Ubuntu and Gentoo have
a command to force threads to run on specific cores?

Ian.

Score
1
June 25, 2010 2:31:36 PM

opmopadopWinzip is dead. Can you use 7zip from now on please Tom!


For benchmark purposes this is true; I second this notion.
Score
0
!