Any Thuban X6 results leaked? [Archive]

Mech0z

03-09-2010, 12:19 PM

It will OC like Phenom II C3 at best. (3.7GHz ~ 4.1GHz with extreme air cooling). If you want more performance, then go after i7-930(it should OC 3.9GHz ~ 4.3GHz with extreme air cooling). It should outperform OC-ed Thuban in most of the real-world apps.

A 930 will cost me 2150kr (420$) while I can get a used 920 3914A743 for 1500kr used. (300$) not sure its worth getting the 930

gOJDO

03-09-2010, 12:20 PM

If the used 920 is rev D0 then go after it. In Q3 this year you'll waste some more unearned money and you'll buy yourself a Gulftown.

charged3800z24

03-09-2010, 03:32 PM

It will OC like Phenom II C3 at best. (3.7GHz ~ 4.1GHz with extreme air cooling). If you want more performance, then go after i7-930(it should OC 3.9GHz ~ 4.3GHz with extreme air cooling). It should outperform OC-ed Thuban in almost all the real-world apps.

How can you be so sure on this? Cause the 2.2-2.6 ghz Opterons don't beat out the higher clocked, turbo boosted Xeons?

accord99

03-09-2010, 04:50 PM

How can you be so sure on this? Cause the 2.2-2.6 ghz Opterons don't beat out the higher clocked, turbo boosted Xeons?
They don't, unless you have twice as many Opteron cores.

tbone8ty

03-09-2010, 04:54 PM

yeah cant wait to see how X6 performs

any reason why 965 has gone up $20 at newegg?

Dumo

03-09-2010, 05:19 PM

No one knows at this moment:D

Probly right now the engraver machine just in the middle of laser etching 1010BPMW Phenom II X6 on on one of the cpu IHS

One of those cpu will snug fit in my mobo's socket and get torture soon

crazydiamond

03-09-2010, 05:33 PM

you mean APMW ;)

SocketMan

03-09-2010, 05:46 PM

We know that Magny Cores is based on the "lisbon" core,and if that is also the
case with Thuban,then OC should be higher then the current Phenom 2 (C3).
AMD cpus are like wine - get better with age or in this case revision ;)

However if you do plan on using all the cores/threads (ex. crunching/folding)
then i7 is the way to go thanks to hyper threading and the lack of any desire,
on AMD's part, to compete for the "high end" (desktop) cpu market.
They could if they wanted too,just need to do the same thing that was done when the 3870 x2 came about,which is basically:Can't beat the competion with
1 gpu ? Why not just add another one,simple and gets the job done.

Dumo

03-09-2010, 05:46 PM

Probly A for Monday, B = Tues etc....:)

So, guys pick your lucky day:D

Since all will be coldbug less anyway....Hope

bingo13

03-09-2010, 05:55 PM

No one knows at this moment:D

I know... :D

Dumo

03-09-2010, 06:13 PM

I know... :DYep...I know that you know:D

Bottom line it will be good....Right?

gOJDO

03-09-2010, 11:52 PM

How can you be so sure on this? Cause the 2.2-2.6 ghz Opterons don't beat out the higher clocked, turbo boosted Xeons?That's one of the reasons and the other is my personal experience with both PII and i7.

Anyway, only few applications can benefit from more than 2 cores, and even if we consider the "megatasking", 4 cores are basically enough unless you use your system for folding/crunching. So, an PII x6 will have advantage of the extra two cores, compared to PII x4, only in some rare situations when CPU is heavily loaded with multithreaded software. But then on the Intel's side, the HT comes into play. With i7's significantly higher IPC, it's performance in applications which can utilize 1,2,3 and 4 cores will be significantly higher than on the PII. And 95% of the desktop software today can't utilizes more than 4 cores.

We know that Magny Cores is based on the "lisbon" core,and if that is also the
case with Thuban,then OC should be higher then the current Phenom 2 (C3).
AMD cpus are like wine - get better with age or in this case revision ;)I doubt this. Having two more cores will put even more heat to the die. And we all know that the primary criteria for successful PII OC-ing are the low temperatures. Also the x6 will have more logic (more transistors that might fail operating at certain freq) which reduces the OC potential.

and the lack of any CPU, on AMD's part, to compete for the "high end" (desktop) cpu market.fixed. :D

They could if they wanted too,just need to do the same thing that was done when the 3870 x2 came about,which is basically:Can't beat the competion with 1 gpu ? Why not just add another one,simple and gets the job done.GPU and CPU tasks have nothing common to compare. AMD currently don't have an architecture and a production process to compete with Intel in the high-end desktop segment. Adding more cores and sockets won't help.

mcmeat51

03-10-2010, 02:55 AM

It will OC like Phenom II C3 at best. (3.7GHz ~ 4.1GHz with extreme air cooling). If you want more performance, then go after i7-930(it should OC 3.9GHz ~ 4.3GHz with extreme air cooling). It should outperform OC-ed Thuban in almost all the real-world apps.

Im intrigued with the gaming performance.
http://www.legionhardware.com/articles_pages/cpu_scaling_with_the_radeon_hd_5970,21.html

AMD phenoms are actually better for gaming when OCed to 3.8+ than a 4Ghz+ OCed i7.

dinos22

03-10-2010, 03:04 AM

I know... :D

^^ yeah sure :p::D

wuttz

03-10-2010, 03:28 AM

Im intrigued with the gaming performance.
http://www.legionhardware.com/articles_pages/cpu_scaling_with_the_radeon_hd_5970,21.html

AMD phenoms are actually better for gaming when OCed to 3.8+ than a 4Ghz+ OCed i7.

and you have to keep in mind intel has hyperthreading, or should i say "double" the number of pipelines feeding the execution units.
because it can actually degrade performance in some applications and is unpredictable in performance across a broad range of software, amd didnt implement it. they rather opted for a pseudo-cluster multithreading architecture.

of course, we'll soon see how it works. but judging from dresdenboy's blog, it looks like its going to be a stunner. :up::up::up:

gOJDO

03-10-2010, 08:10 AM

Im intrigued with the gaming performance.
http://www.legionhardware.com/articles_pages/cpu_scaling_with_the_radeon_hd_5970,21.html

AMD phenoms are actually better for gaming when OCed to 3.8+ than a 4Ghz+ OCed i7.Actually they are not, according to the link you provided. 4GHz i7 vs 4GHz PII:
Batman: 106.3%
Wolfenstein: 98.9%
Left 4 Dead: 103.1%
COD: 97.8%
CoH: Oposing Fronts: 106.9%
Crysis Warhead: 100%
FarCry 2: 94.2%
Battle Forge: 97.9%
H.A.W.K: 96.5%

In average the i7 @4GHz has 0.18% higher average FPS than the PII @4GHz @2560x1600 resolution with high AA/AF, or basically they are identical when both clocked to 4GHz (although the NB/L3 clocks are not mentioned). BTW, average FPS is not the only factor for the good gaming experience. What matters the most is the min FPS. ;)

To complete the image, I suggest to check xbitlabs review. It is more comprehensive(has lowest and average FPS) with more benchmarks at different settings (resolutions and AA/AF), including single 5870 and x-fire 5870: http://www.xbitlabs.com/articles/video/display/radeon-hd5870-cpu-scaling_4.html#sect0

You can notice that at lower resolution like 1280x1024 with a single 5870, or when the GPU bottleneck is diminished, the i7 @4.1GHz outperforms the PII @4.1GHz in average. But when you add the second 5870 in the system, then i7 even at 2.67GHz outperforms a 4.1GHz PII.

charged3800z24

03-10-2010, 08:36 AM

Actually they are not, according to the link you provided. 4GHz i7 vs 4GHz PII:
Batman: 106.3%
Wolfenstein: 98.9%
Left 4 Dead: 103.1%
COD: 97.8%
CoH: Oposing Fronts: 106.9%
Crysis Warhead: 100%
FarCry 2: 94.2%
Battle Forge: 97.9%
H.A.W.K: 96.5%

In average the i7 @4GHz has 0.18% higher average FPS than the PII @4GHz @2560x1600 resolution with high AA/AF, or basically they are identical when both clocked to 4GHz (although the NB/L3 clocks are not mentioned). BTW, average FPS is not the only factor for the good gaming experience. What matters the most is the min FPS. ;)

To complete the image, I suggest to check xbitlabs review. It is more comprehensive(has lowest and average FPS) with more benchmarks at different settings (resolutions and AA/AF), including single 5870 and x-fire 5870: http://www.xbitlabs.com/articles/video/display/radeon-hd5870-cpu-scaling_4.html#sect0

You can notice that at lower resolution like 1280x1024 with a single 5870, or when the GPU bottleneck is diminished, the i7 @4.1GHz outperforms the PII @4.1GHz in average. But when you add the second 5870 in the system, then i7 even at 2.67GHz outperforms a 4.1GHz PII.

at 2.6ghz, is turbo on? cause then it isn't running 2.6ghz.

gOJDO

03-10-2010, 08:52 AM

http://www.xbitlabs.com/images/video/radeon-hd5870-cpu-scaling/01_i7-2.67_cpu-z_big.png1
http://www.xbitlabs.com/articles/video/display/radeon-hd5870-cpu-scaling_2.html
Turbo is off, else CPU-z will show x21 multiplier and 2.8GHz instead of x20 multi and 2.67GHz.

wuttz

03-10-2010, 08:52 AM

You can notice that at lower resolution like 1280x1024 with a single 5870, or when the GPU bottleneck is diminished, the i7 @4.1GHz outperforms the PII @4.1GHz in average. But when you add the second 5870 in the system, then i7 even at 2.67GHz outperforms a 4.1GHz PII.

if a gamer has a 5870 and plays at 1280x1024, then something is totally wrong with him. and doubly so if he crossfires it and still plays at the same resolution.

1080P is the minimum resolution i would consider as "actual gaming." anything else is for marketing.

gOJDO

03-10-2010, 08:58 AM

if a gamer has a 5870 and plays at 1280x1024, then something is totally wrong with him. and doubly so if he crossfires it and still plays at the same resolution.

1080P is the minimum resolution i would consider as "actual gaming." anything else is for marketing.I absolutely agree with you.

1280x1024 is only used for the purpose of CPU performance benchmark. At lower frequency the graphics cards can produce more FPS, thus they put more overhead to the CPU. If, somehow, the graphics cards bottleneck can be avoid then you'll be able to benchmark the CPU(and memory subsystem) gaming performance. That's why CPU gaming benchmarks are being performed at lowest possible resolutions and settings.

In reality there is NO difference in gaming experience between a C2Q, Ci7 and PII clocked at 4GHz with an OC-ed 5870(1000MHz/1300MHz) playing on 1920x1200. I didn't notice any difference and I doubt that a human being can notice.

Glow9

03-10-2010, 09:26 AM

Majority of games still don't take advantage of multiple cores but they are getting there. So at this point this thing is just gonna be bling, kinda funny discussion. I've always though hyperthreading was a gimmick considering our Athlon XPs could hold their own against the more expensive P4 w/Hyperthreading. Clock cycles for the win.

demonkevy666

03-10-2010, 09:36 AM

demonkevy666

03-10-2010, 10:04 AM

Actually they are not, according to the link you provided. 4GHz i7 vs 4GHz PII:
Batman: 106.3%
Wolfenstein: 98.9%
Left 4 Dead: 103.1%
COD: 97.8%
CoH: Oposing Fronts: 106.9%
Crysis Warhead: 100%
FarCry 2: 94.2%
Battle Forge: 97.9%
H.A.W.K: 96.5%

In average the i7 @4GHz has 0.18% higher average FPS than the PII @4GHz @2560x1600 resolution with high AA/AF, or basically they are identical when both clocked to 4GHz (although the NB/L3 clocks are not mentioned). BTW, average FPS is not the only factor for the good gaming experience. What matters the most is the min FPS. ;)

To complete the image, I suggest to check xbitlabs review. It is more comprehensive(has lowest and average FPS) with more benchmarks at different settings (resolutions and AA/AF), including single 5870 and x-fire 5870: http://www.xbitlabs.com/articles/video/display/radeon-hd5870-cpu-scaling_4.html#sect0

You can notice that at lower resolution like 1280x1024 with a single 5870, or when the GPU bottleneck is diminished, the i7 @4.1GHz outperforms the PII @4.1GHz in average. But when you add the second 5870 in the system, then i7 even at 2.67GHz outperforms a 4.1GHz PII.

those test where done with 5970 not 5870

and your other test show DDR2 on Phenom II
ram is only clocked to 908mhz too.
Nb is 227 x 10 = 2270 is a bottleneck @ 4.1ghz Phenom II

I7 there has 1600mhz tri channel ddr3 vs 908mhz ddr2 and 1092 ddr2. dual channel
though I'm sure how much ram can change game play on some those games.
I know my gta4 hates have only 2gbs must have 4gb to get good play, but I'm only using junky 3870's >_<

but I see one core is about comparable to 800mhz clock speed so 3.0ghz six core = to about 4.6ghz 4 core for phenom II

mcmeat51

03-10-2010, 10:52 AM

To complete the image, I suggest to check xbitlabs review. It is more comprehensive(has lowest and average FPS) with more benchmarks at different settings (resolutions and AA/AF), including single 5870 and x-fire 5870: http://www.xbitlabs.com/articles/vid...g_4.html#sect0

I do see your point, however.

This is not completely fair as its using an AM2+ board and DDR2. Also this AMD board is not even one of the better ones for gaming after looking at some comparisons, in fact its mediocre at best.
The Intel P6T however is a great board for gaming amongst the 1366 boards, near the top of its league.

If someone can show me some comparisons between a decent AM3 board such as the MSI Gd70 or the ASUS crosshair with a 965 @ 4Ghz 2.6NB against an 920 @4Gz with the same speed ram then Ill be more then that would convince me. oh and using the latest 10.3 catalyst drivers

as far as bang for buck AMD I think is far more appealing, perhaps if I had a grand to spare I may get an i7 system.

accord99

03-10-2010, 11:15 AM

Legion hardware has another CPU scaling review up:

http://www.legionhardware.com/articles_pages/radeon_hd_5870_crossfire_cpu_scaling_performance_p art_1,1.html

tbone8ty

03-10-2010, 11:20 AM

Amd should be bundling ln2 kits with there chips then will see how she scales at 6-7ghz muahaha

gOJDO

03-10-2010, 12:12 PM

Legion hardware has another CPU scaling review up:

http://www.legionhardware.com/articles_pages/radeon_hd_5870_crossfire_cpu_scaling_performance_p art_1,1.htmlVery nice find. In the 1st article, they didn't described their testing methodology, but in the 2nd they summed it up nicely:

However this time around we are testing a pair of Radeon HD 5870 graphics cards without enabling any AA/AF quality settings. Furthermore, we are not just testing at 2560x1600, and will instead include a low resolution test using 1680x1050. This will allow the high-end processors to really stretch their legs, and will uncover which processor would provide the best performance assuming no limits.

Their systems are now using the same components except the mainboards and the CPUs, same operating systems and same drivers. I summed up their scores in an excel sheet, inorder to put the advantage of Core i7 @4GHz over PII x4 @4GHz in a single table:
http://img169.imageshack.us/img169/6820/i7vspii.jpg

Considering all the benchmarks between the i7 @4GHz and the PII X4 @4GHz, the i7 is 28.4% faster in average. In my book, that would mean that the 4GHz i7-9xx is mopping the floor(especially at min FPS which IMO is the most important factor for fluent gameplay) with the 4GHz PII x4 9xx. The PII x6 should reduce the gap between them, but IMO only a little.

So PII x6 would be a good choice for very few apps that can utilize its 6 cores, but we'll have to see it's pricing before we can assume if it'd be worth the money. For general purpose software and general users, PII x4 will do it's job better than a PII x6 due to it's higher core clocks.

Hell Hound

03-10-2010, 12:29 PM

lol a 3.4 x6 should nip @ i7 975 :)

Mats

03-10-2010, 01:13 PM

I don't think the X6 will help much in current games. Higher clocks or IPC is needed, not more cores.

mcmeat51

03-10-2010, 01:45 PM

ok fair enough, slightly annoyed at the i7's lead lol.

wuttz

03-10-2010, 01:47 PM

Considering all the benchmarks between the i7 @4GHz and the PII X4 @4GHz, the i7 is 28.4% faster in average. In my book, that would mean that the 4GHz i7-9xx is mopping the floor(especially at min FPS which IMO is the most important factor for fluent gameplay) with the 4GHz PII x4 9xx.

no AA/AF, and lower the resolution to "stretch the legs," or to benchmarket the core i7 over the Phenom II, in simple words.

you have to consider the games they included in that line-up were mostly the games that won over the Phenom II in their 5970 review. :shrug:

in their 5970 review, the Phenom won in:

wolfenstein, battle forge, hawx, COD: MW2, far cry

and lost in:

coh, batman, left4dead 2,

and tied in:

crysis

all within few fps' of each other.

but, thats 5 wins for the phenom to core i7's 3 and tied at crysis. thats hardly mopping the floor at actual game settings and resolutions, unless again you play with no AF/AA on with a 5870/5970.

now to make my point clear.

in actual, realistic, gaming set-ups, you will see no difference in a phenom II or core i7's performance, unless you buy a 5870 and game at 1280x1024, in which case i suggest rather getting a 4890 instead and at least a 1900x1200 or 2048x1152 display.

with amd's phenom 2, you spend less on your platform leaving you enough money for an extra LCD, another video card, a better psu, or just extra budget for anything else. when you consider the prices for the equivalent intel set-up, you have to spend for a more expensive cpu, a more expensive motherboard and an extra stick of ddram- and yet you will see just the same performance if you made an amd pc(unless you really notice the difference between 110fps and 100fps.)

but somehow, intel has to justify you spending more for their hardware so they need to really "stretch their benchmarketing" on this one.

keithlm

03-10-2010, 02:45 PM

Those are the worst graphs I've seen in awhile.

How can you add min and average FPS and then show a system with a higher average FPS appear lower in the graph?

Ugly. And not very usable.

charged3800z24

03-10-2010, 05:59 PM

Very nice find. In the 1st article, they didn't described their testing methodology, but in the 2nd they summed it up nicely:

Their systems are now using the same components except the mainboards and the CPUs, same operating systems and same drivers. I summed up their scores in an excel sheet, inorder to put the advantage of Core i7 @4GHz over PII x4 @4GHz in a single table:

Considering all the benchmarks between the i7 @4GHz and the PII X4 @4GHz, the i7 is 28.4% faster in average. In my book, that would mean that the 4GHz i7-9xx is mopping the floor(especially at min FPS which IMO is the most important factor for fluent gameplay) with the 4GHz PII x4 9xx. The PII x6 should reduce the gap between them, but IMO only a little.

So PII x6 would be a good choice for very few apps that can utilize its 6 cores, but we'll have to see it's pricing before we can assume if it'd be worth the money. For general purpose software and general users, PII x4 will do it's job better than a PII x6 due to it's higher core clocks.

That link above has some weird issues with the results. How in the heck does thew Intel chips have the same FPS but still plot farther on the graph? Something is weird.

Mats

03-10-2010, 06:03 PM

That link above has some weird issues with the results. How in the heck does thew Intel chips have the same FPS but still plot farther on the graph? Something is weird.
Each bar is a combination of min and avg FPS. Not min bar in front of avg bar.

charged3800z24

03-10-2010, 06:23 PM

Each bar is a combination of min and avg FPS. Not min bar in front of avg bar.

I notice that now, still looks bad if someone is just glancing over it. I found another strange thing. Look at this page here.

http://www.legionhardware.com/articles_pages/radeon_hd_5870_crossfire_cpu_scaling_performance_p art_1,15.html

Did anyone else notice the Phenom II rig Speed up when the resolution was increased. and the I7 slowed down? You think the guy doing the artical would have mention this. The 4.0ghz Phenom II did 137FPS average @ 1680 res. at the 2560 res it did 146.5 FPS. The I7 actually dropped performance. Lower res is supose to eliminate GPU bottle neck, how does this happen on the inferior AMD setup?

BeepBeep2

03-10-2010, 06:50 PM

Thats so sad...the review is just screwed up to make the AMD CPU look weaker?

SocketMan

03-10-2010, 06:52 PM

GPU and CPU tasks have nothing common to compare. AMD currently don't have an architecture and a production process to compete with Intel in the high-end desktop segment. Adding more cores and sockets won't help.

Gaming is a common task for cpu & gpu.Phenom 2 is a better over clocker then
a dual core K8 or the 1st phenom - both have less "logic" then the PH2.
More sockets and cores won't help with supepie, as for other things, a picture says a thousand words:

FlanK3r

03-11-2010, 01:48 AM

Phenom II is "strange"in games resolutin, it dropps not so much as example Core Quad or i7 with higher and higher resolutions. This had too old Phenom I in games. Charged show it in link...But back to Thuban guys? No war here AMD vs Intel yet :)

gOJDO

03-11-2010, 03:38 AM

Phenom 2 is a better over clocker then
a dual core K8 or the 1st phenom I agree that it's better OC-er than the 1st Phenom, but I disagree that it's better than K8. K8 was much more OC able. For example, back in the K8 days I had an Opteron 175 OC-ed to 3GHz stable 24/7. Today I have 3 Phenom's and none of them work stable OC-ed for the same %. Just check how many people have their G2 K8 dualcores OC-ed for more than 30%. Then check how many have pushed their Phenom II x4 over 30%.

- both have less "logic" then the PH2.And both are made on older, worse performing production process, while the one has different architecture optimized for a 3 generations older production process.

A new revision (rev. C4) should bring some improvements regarding to clocks and voltages (read power consumption and temperatures). But the adding ~30% more logic will have negative effects on the frequency side, as well as power consumption and temperatures. That's why I doubt that it'll OC better than the current PII x4. Actually looking at the frequencies of the server versions, I don't belive it'll match PII X4.

informal

03-11-2010, 04:55 AM

Less speculation and more facts please. C3 ,especially in 555BE version, is THE best buy CPU at the moment from perf./price POV (and this is only counting in the CPU factor,not to mention the board factor).
Thuban will be a pretty good desktop MPU and it will be priced very aggressively.No SMT gimmick will make i7 better in true MT workloads,it simply won't fly against a real 10h cores (50% more than in QC Denebs). In less MT aware desktop workloads the difference between Deneb and Nehalem is not so pronounced that it should make someone cache out for additional board coast and tripple channel ram. Just OC your Phenom II and you are good to go since 15% won't make any difference to 99% of desktop buyers.

snoro

03-11-2010, 06:06 AM

i dont think thuban is aimed at pure gamer, the phenom II x4 955 965 can do that job perfectly, Thuban will be more for people who mix stuff with their computer, they wont just play with it or just overclock with it but they will also use the extra core for crunching/folding and multi tasking or rendering, true target consumer of the thuban will do a big mix of all that. Thats why they will buy the 2 extra core.

crazydiamond

03-11-2010, 09:30 AM

Less speculation and more facts please.

took the words out of my mouth :clap:

FlanK3r

03-11-2010, 09:49 AM

2 Thubans promotion slides (http://translate.googleusercontent.com/translate_c?hl=de&ie=UTF-8&sl=de&tl=en&u=http://news.ati-forum.de/index.php/news/35-amd-prozessoren/1111-exklusiv-thuban-kommt-als-phenom-ii-x6-1000-serie&prev=_t&rurl=translate.google.com&usg=ALkJrhh1ALJ9kWzpyrJcxF8k7k4TdgP_mQ)

http://news.ati-forum.de/images/stories/Szymanski/News/2010/thuban_phenom_x6_1000series_1.jpg

Lagg Master

03-12-2010, 06:32 PM

Anyone think the 960T(Zosma) will be a 6 core with 2 cores locked?

FlanK3r

03-13-2010, 12:46 AM

yes...but u need update BIOS for Zosma and in new BIOS are ACC disabled think ,-)

Lagg Master

03-13-2010, 09:42 AM

yes...but u need update BIOS for Zosma and in new BIOS are ACC disabled think ,-)

Asus is releasing motherboards with core_unlocker switches on them. This is one of the reasons why i think i will go for the 960T. From what i have read it is basically the 955 with an extra 100MHz speed. It will have the C4 stepping that helps boost speeds on single threaded programs by shutting all cores down except one and overclocking that core.