Printable View
http://translate.google.com/translat...n&hl=&ie=UTF-8
no software required, no bios update required... it just adjust the multiplier based on how many cores are loaded... same as intel it seems, cool :)
i wonder if itll drop out of turbo mode if current draw or temps are too high like intel chips... they didnt test that so far...
expreview
i prefer the non translated version
lolz
Voltage jumps pretty high when in single-core turbo mode.
or cpu-z doesn't read correct to start with or they modified the bios, no accurate up-to-date bios..... a bit to soon for conclusions i thinkQuote:
Originally Posted by richierich
Core VID is not actual voltage applied, even AMDs own Overdrive likes to show VID instead of actual voltage in many places. HOWEVER odd enough, if you see i think "VCore", then CPUz is reading the right voltage. i cant understand it, but for me, half the time i start up CPUz, it shows one or the other, but its always correct, its just a question of if its reading what we want to see, lol
Very nice implementation :)
Seems it scored close to i7 860 which is very positive news, this will score above i7 920 for sure but with the i7 930 it will be close...
In general Lynnfield does not perform as well as a Bloomfield clock 2 clock, given that $10 separate i7 930 and i7 860 its a easy choice.
I5 750 vs 1035T will be great and epic fight and i dont think that the i5 750 will survive because with no HT it cant really do that good in bench's....!!!
[QUOTE=ajaidev;4335978]
Seems it scored close to i7 860 which is very positive news, this will score above i7 920 for sure but with the i7 930 it will be close...
In general Lynnfield does not perform as well as a Bloomfield clock 2 clock,[QUOTE]
http://prohardver.hu/dl/cnt/2009-09/...results/cb.png
from 1.4000 to 1.475v is a big jump? :confused:Quote:
Originally Posted by richierich
1055T on XS :D
http://www.xtremesystems.org/forums/....php?p=4335871
[QUOTE=Oliverda;4335996][QUOTE=ajaidev;4335978]
Seems it scored close to i7 860 which is very positive news, this will score above i7 920 for sure but with the i7 930 it will be close...
In general Lynnfield does not perform as well as a Bloomfield clock 2 clock,Yep seems the i7 920 is out of the picture :D and since the 1055T is suppose to cost quite a bit less than $300 this is great.
i7 930 is its main enemy but its priced at $290-$300 where the 1090T will be so this is great. I am sure that a 1090T will beat the crap out of i7 930 due to the clock difference and the closed thing to that clock speed is the i7 870 and the i7 950 which are for $550 and $570.
This is gonna be a very interesting/Bloodbath if the rumored prices are true "1090T for $300, etc"
Here is 1055T vs 1090T
http://www.expreview.com/img/news/20...55t/cin11b.jpg
http://item.slide.com/r/1/237/i/vGMr...knwu6fLx9AYMD/
As i said above i7 870 and the i7 950 which are for $550 and $570.
1090T score in C11.5 shows some odd scaling going from 2.8Ghz model,only 6% difference while the clock difference is double of that percentage wise. Could be some bios issue with CnQ
It seems everytime I check back, something new pops up about turbo core that makes it seem rushed and poorly implemented. The biggest deal-breaker (IF TRUE; check out "Turbo CORE 101" thread) is that during turbo boost, all the idle cores are still sucking the same juice as active cores! That's dropping the ball right there! That means even in single core turbo situations, 5 idle cores would be sucking as much as 1.5v+ while doing nothing.Quote:
Originally Posted by saaya
IF ALL THESE TURN OUT TO BE TRUE, then, IMHO AMD should not have implemented turbo in the first place. It makes them look more like amateurish copycats who would rush a badly implemented feature to market just to score some few benchmarking points. The x6 line is already a very good bargain (compared to Intel offerings) without the Turbo CORE feature! And why not just call it Turbo Boost? I mean the cores are boosted during this stage and not CORED or anything :lol: Regardless of nomenclature, I'm sure Intel still gets to collect their tax. So all in all, very tax-unfriendly, and environmentally unfriendly implementation. But I'd still get it for a dedicated encoding system, unless Intel brings an affordable hexacores to market some time soon.
The cannot suck "the same juice as the active cores" since they run at 800Mhz... The power draw of those idle cores with turboed Vcore will be very small. The cores won't do anything,they are called idle for a reason :rolleyes:
They dont really suck the same amount to power. Their speed is lowered a lot and are in idle mode. The 1.475v is for all the core 3 idle and 3 in turbo that does not mean that the idle ones will eat the same amount of juice as the turbo once because there is no way in hell AMD could have fit that situation in the TDP listed....Quote:
Originally Posted by OhNoes!
You can rolleyes all you want, but I'm bringing up something you have conveniently chosen to ignore. Also, looking at your posting history, I wouldn't take anything you say or believe about hardware seriously. Since you have in the near past gloated openly over Dresdenboy's blogs and 'findings' here is something he posted in the "Turbo CORE 101" thread with regard to Turbo CORE implementation in THUBAN:Quote:
Originally Posted by informal
http://www.xtremesystems.org/forums/...&postcount=119Quote:
Originally Posted by Dresdenboy
What says you to that?
64bit test (1055T on 890GX and supported bios from expreview.com) vs 32bit (1090T on 790gx and unsupported bios from Coolaler's forum)Quote:
Originally Posted by ajaidev
Informal isn't ignoring that, he understands it. He's being pretty clear, maybe you need a lesson in reading comprehension? The idle cores are IDLE. Test it yourself. Idle your CPU. Then increase the voltage, and let it idle. You wont see much of a jump in power usage. It'll be even less pronounced at lower clockspeeds.Quote:
Originally Posted by OhNoes!
It dosn't matter what freq they run at in idle. All the consumption is comming from a leakage. So it depends how well it leakage staedy since the power consumption is proportional to the square of voltage.Quote:
Originally Posted by informal
Only it's not really idle, it's still eating power - power is NOT COMPLETELY SWITCHED OFF! Read the quote from Dresdenboy.... he clealy described it as a quote "problem." And what's so elegant about applying voltage across idle and turboed cores?Quote:
Originally Posted by ryboto
since when does idle mean switched off?????? It's only a problem in the sense that Intel can power down the cores, and AMD can't at the moment. It's just a competitive problem.Quote:
Originally Posted by OhNoes!
you guys give it up, the king has spoksen. OhNoes! says thuban is :banana::banana::banana::banana:, then thuban must be :banana::banana::banana::banana:.
It's proportional at a given frequency. When I undervolt my CPU at 1ghz, I only save 3-4W. When I undervolt at 3.0ghz, I save 10-15W. This isn't as big an issue as you guys want it to be. It doesn't best Intels power saving techniques, but the fact that even with turbo core enabled the CPU can remain under a 125W TDP is still impressive for 45nm.Quote:
Originally Posted by kl0012
Switched off, powered down, what's the difference? The thing is the cores are not powered down (no power gating) and power could still be dissipated through leakage. Very inefficient and crude implementation at best.Quote:
Originally Posted by ryboto
I have already said that the x6 line is good, and that they don't need turbo to compete. I guess fanboys like you are just averse to any type or form of criticism from any quarter. If even people like Dresdenboy think it's a problem, then why can't I say AMD could have done better without TC implementation at all, or improved the current implementation?Quote:
Originally Posted by crazydiamond
kl0012 is correct that it's the leakage power that may be the problem,but with Thuban there should be some improvements when it comes to 45nm SOI process . What intel can do with Nehalem is power gating at the each core level which is better approach than AMD's (Llano and BD cores will have this next year though)
BTW OhNoes,nobody ever took you seriously ,with the last posts you just cemented that sentiment.:p:
Well he somehow managed to not understand it... He may as well test it with his current system and see the difference himself.Quote:
Originally Posted by ryboto
Yep. It proportional to frequency too, but in idle when a core is in idle a frequency is less important since most of transistors dont switch. But the leakage is always proportional to the square of voltage. It just a physics (P = U^2/R).Quote:
Originally Posted by ryboto
I didn't say it bad. I said it depends.Quote:
This isn't as big an issue as you guys want it to be.
XS just got one retail 1055T in AMD section. User imamage got one in retail store :
http://www.xtremesystems.org/forums/...d.php?t=249434
Also there is one 1090T in AMD section as well!
power gating per core should make some incredible laptop cpus, where every watt is really noticed.
@ Informal: Finally, you admit there's a problem with the implementation, even if while being sour and sulky at the same time! I don't know how you do it, dude.
Ok, so they can't power down their cores, so they lack power gating of individual cores...so what? Like I said, if the CPU can still stay in the rated TDP during turbo core boosts, what's the problem?Quote:
Originally Posted by OhNoes!
Why are you describing it as a problem? It's not as advanced as core gating, but it's going to work problem free and within the rated TDPs, so why are you focusing on it? Anyone following this news already knows that about the x6's, so just drop it.Quote:
Originally Posted by OhNoes!
ok so they add turbo to a cpu that works on >2 year old motherboards, and u still think theres problems? or how about they did it without having to start from scratch.Quote:
Originally Posted by OhNoes!
they just gave phenom a new feature, and are pricing it well within anyones reach, but thats just not enough?
Oh noees!, Houston we have a problem!!! /panic mode off :rolleyes:Quote:
Originally Posted by OhNoes!
First ,the "problem" is not a big issue since cores are in low clock mode,they do draw power even then but not nearly as much as you would think. Second , "problem" becomes and advantage when you do not need to have some additional circuitry inside your chip to average the temp. readings in your MPU in order to know if your chip can actually use Turbo or not. With AMD there is one htc limit temp. than is the limit for any chip they make. Turbo CORE will engage always and won't depend on the cooling solution or cramped case one is using. Even with the def. heatsink and fan it will engage if one condition is met : OS is not using half of the cores.
could it be better, sure, everything has room for improvement, if it didnt we would never get new hardware. to say that they would have been better off with no turbo is just hating. 3.3ghz for poorly threaded apps is better than 2.8ghz no matter how you slice the bread. if they stay under tdp who cares how they implemented it. just let it go man, let people be happy about the new hardware - no reason dump all over amd because its not how YOU would have done itQuote:
Originally Posted by OhNoes!
edit: lol i see the last 3 posts all pretty much saying the same thing. hopefully we can move along now.
sorry but i have this bad taste in my mouth now and nothing can make me hap... ooooo look at the pretty benchmarks:p:Quote:
Originally Posted by crazydiamond
I just wanted to point out one thing.
AMD's 3 core Turbo core in 1055T pushes 3 cores 500mhz more.
Intel's Turbo Boost pushes 1 core 540mhz more in i5 750 and 660mhz more in i7 860
Finally Intel's Turbo Boost pushes 1 core just 260mhz in case of Bloomfield i7 930.
The difference in turbo is so much that a i7 870 can out perform a i7 950 in some cases. 1055T on the other hand is very different and unlike the Intel ones the dual state only process will help it greatly in situations that require two cores.
While two cores are active 1090T does 3.6Ghz, 1055T does 3.2Ghz, i7 870 does 3.4Ghz, i7 860 does 3.3Ghz, i7 930 does 3Ghz and i5 750 does 3.2Ghz
EDIT: Forgot to include the freq if 3 cores are active instead of 2. i7 870 does 3.2Ghz , i7 860 does 2.9Ghz , i5 750 does 2.8Ghz , 1055T does 3.2Ghz and 1090T does 3.6Ghz.
Thus the true power of this type of a turbo lies in dual/triple threaded application. Just imagine a 1090T running at 3.6Ghz @ 3 cores while i7 870 runs at 3.2Ghz @ 3 cores, difference between the two is 400mhz and by the looks of it if a application uses three threads it will be very beneficial for these processors.
Yes, that is quite a lot.Quote:
Originally Posted by saaya
<-------- hope these chips will give Intel a reason to rethink their pricing strategy.
VID and voltage should not be mixed up. i dont think that a .1v increase to VID means .1v increase to core voltage. i think the VID exists so you can fine tune the voltage. but im not chew or informal, or one of the other AMD experts, so please dont assume im right.Quote:
Originally Posted by zalbard
according to Dresdenboy, its 1.225 stock, and 1.425 when Turbod. these values might be different for each model though, i expect the 95W chips to be like 1.2v and 1.35v
http://img103.imageshack.us/img103/1...tmapimagec.png
I did not include core 5 and 6 because it same as the 4 core for the Thuban.
ajaidev: u compared 32-bit vs 64-bit...64-bit is more better clock for clock
^ ya i know did not see it when i posted, must be the spree of the moment :D
if were talking about CB11.5, on my 940 its like 3.35 vs 3.5 (i think, i did it yesterday and im at work so i cant tell, but i think the difference is like 5%, unlike CB10 which is 25%)Quote:
Originally Posted by ajaidev
omg...now bit OT, but think Bulldozer will terror of Intel :), AMD going more and more UP
well, better this than no turbo core at all imo...Quote:
Originally Posted by OhNoes!
but yeah, it does seem rushed and for mobile cpus and power conscious users its not a good feature... and in servers it might not be that great either, thats where turbo boost also seems to make a lot of sense as it reduces idle power consumption... but doesnt amd have at least 2 power planes on their cpus? so if half the cores or more are idle im sure they at least reduce vcore if not even shut it off?
and the idle cores will consume power and wont be turned off like on intel cpus, which means higher power consumption for the amd cpu, right? well overall they consume about the same power... intel uses the power more efficiently, but in the end what matters is overall power consumption and performance... if amd can be competitive there with a worse implementation then who cares?
To everyone complaining about lack of Power Gating:
Power Gating requires special type of transistors which are not available on GloFo 45nm process. Therefore AMD couldn't design CPU on this process with Core Gating technology. This is planned for 32nm process.
One thing to remember is that AMD decided to introduce Turbo Mode only using improved 45nm process from GloFo. They could easily copy Turbo even with Rev. C3 of Phenoms but they decided to wait for Lower K dielectric which limits leakage.
I hope this clears things up a bit.
So it should be available in two weeks... I can't wait! I want benchmarks before that...
did amd can lock at turbo multi
like intel board?
This will probably be the first AMD cpu I'll buy for my main rig after 4 years or so.
its a 9% boost in voltage... hows that a lot? or do you mean 1.475v is high? on the latter i agree, but a 9% boost in voltage isnt much imo...Quote:
Originally Posted by zalbard
no, intel has a turbo multiplier even for all cores active... i dont think amd does.Quote:
Originally Posted by PatRaceTin
intel only did that cause if they would run the cpus at the 4core turbo all the time, they would be above the 125W tdp :D
power consumption is powered (^), so 9% is really like 20% more power draw. to put it in perspective, with CnQ, if i have all 4 cores at 1.0v 800mhz maxed out load, it still consumes less power than when at 3.0ghz 1.35v idling. boosting that to 1.5v when i was water cooling, i think it drew over 100W more, from the wall, so like 65-70W more on the cpuQuote:
Originally Posted by saaya
With results like these. I can see Intel i7/i5 price drop soon. ^_^
According to here :
http://www.fudzilla.com/content/view/18436/1/
http://forum.coolaler.com/showpost.p...6&postcount=25
Time for amd has come here amd should get out Athlon X6 to compete core i5 750 or something like Athlon 630
So a i7 over 3.4 GHZ will beat a Phenom X6 at 4.2 ghz eh? Well I'm back to buying an intel rig. :)
Head over to the AMD section to check out some better benchies.Quote:
Originally Posted by vitchilo
If you look closer the user disabled 2 cores in order to get to 4.2Ghz ,it's clearly seen in Everest benchmarks... Why do people overreact and jump to conclusions without looking closer?Quote:
Originally Posted by vitchilo
looks like 4.2ghz i7 is just a hair below 4.2ghz phenom x6Quote:
Originally Posted by vitchilo
maybe u should be back to buying amd rig ;) might save a few bucks too ;)
http://hostimage.co.uk/08-Apr-10_0F69-4BBDB0E5.jpg
http://i70.photobucket.com/albums/i82/mrjohnson2/vv.jpg
thought id post it in here instead of creating a new thread ... New 6-core AMD CPUs overclock to 4GHz on air
From what we've seen in the secret testing labs here in Taipei, that might well be the case. We got the chance to see one of AMD’s upcoming 6-core desktop CPUs being overclocked, and the chip hit a whopping 4 to 4.2GHz with a very good air cooler.
That said, this type of overclocking will be limited to Black Edition models, due to HyperTransport limits. As with the original 65nm Phenoms, these new 6-core CPUs will hit around 230MHz on the HT clock before refusing to go any further - so you'll need to use multiplier overclocking, which requires the unlocked multiplier you get with a Black Edition CPU.
if its already been posted soz guys
http://www.bit-tech.net/news/hardwar...-4ghz-on-air/1
6 thread AMD vs 8 thread i7 at same speed and AMD wins (cinebench multithread test) .. nuiiiice . :up:
@alfaunits : you're right , changed :)
"HyperTransport limits"Quote:
Originally Posted by jazkat
what limit.. ? I'm only a where of the HT 2600mhz limit cause instability above that.
http://www.madshrimps.be/?action=getarticle&articID=974
http://www.madshrimps.be/articles/AM...oofd-29959.jpg
chew* said there is a 400mhz wall on AM3
cold scaling is always better with these chips
the chips loves cold and as always with these chips the hit a wall 400mhz on cold for HTT Ocer look at some sempron 140 on Hwbot.
@Criminal, That's 8 thread, not 8 core i7.
balancing the HT and the NB speeds have been a pain, its possible that one of those was too high which seemed like a 230mhz HTT limit, but was something else with a multi too high.
weve already seen someone hit 240mhz on here without a problem.
Those bit-tech guys have no need to look at Far East,we have guys from there posting retail chip results right here @ XS and there is no wall @ 230Mhz for HTT... Actually here user ridney managed 250Mhz HTT with 1055T(locked,14x multi),stable enough to run Cinebench 10 64bit with Turbo core ON(kicked single thread perf. level to 4.1Ghz!)Quote:
Originally Posted by jazkat
hi sorry i didnt know there was guys already testing thuban im a newbQuote:
Originally Posted by informal
thanks for the link its all very interesting.
Someone is chuffed that a Phenom II X6 1055T sample turned up in their labs.
They were keen to test ASUS's claim that even bottom end boards would work with these CPUs and promptly set up a test bench with their favourite budget AM3 board, M4A78LT-M LE, a great budget AM3 board, just needs OC Profile to be perfect and very popular for unlocking X2 550s.
First POST failed but after a BIOS flash to 602 and the board was able to run the CPU.
The system proved to be LinX 20 stable at stock speeds an overclock was applied using a 243MHz FSB resulting in ~3.4GHz on a BE Stock Cooler. The board under test usually tops out around 260-270MHz so there may be more headroom.
Bit-Tech's source: http://forum.coolaler.com/showpost.p...6&postcount=25Quote:
Originally Posted by jazkat
Sorry if this has been posted before:
http://valid.canardpc.com/show_oc.php?id=1132113
Thuban+890GX+Ln2 = 6.29Ghz validated.
holy voltage!
Oh man this validate killed me!!!
1095t?Quote:
Originally Posted by informal
Maybe a glitch in CPUz database? Dunno really.Quote:
Originally Posted by Oliverda
With voltage that high, I suddenly need to see a CPU that runs at 2 volts. Who cares about clocks anymore, pump the voltage! :shocked::shocked::shocked:Quote:
Originally Posted by richierich
yes, its CPU-Z wrong name, in real was 1090T