Sign in with
Sign up | Sign in
Your question

AMD pulls a rabbit out of their Agena hat

Tags:
Last response: in CPUs
Share
November 1, 2007 1:41:46 PM

Yes folks, my prayers have been answered. And AMD has an answer to Intel's single threaded prowess, sort of. What's he babbling about you say? Well, a few months ago I thought that with the split planes and individual core voltage it would be possible for AMD or an OEM to create a utility that would allow you to adjust the frequency of a single core while turning the others down.
Well, according to the folks over at Expreview, they have such a utility being readied for the Phenom launch. Here is a pic of the utility.


So that means you don't have to run all four cores at 3.2GHz, you can just run on ethat high and keep the others at their default or lower speed. Tis will increase single-threaded performance and keep power down. I guess this is what they were talking about in the article a few days ago.

I was oping they'd say they got a retail chip and it's 10-15% faster than the ES. Still good news for AMD enthusiasts though.
November 1, 2007 1:57:48 PM

A rabbit?
I was hoping theyd finally pull some processors out of it.
November 1, 2007 2:22:27 PM

BaronMatrix said:
Yes folks, my prayers have been answered. And AMD has an answer to Intel's single threaded prowess, sort of. What's he babbling about you say? Well, a few months ago I thought that with the split planes and individual core voltage it would be possible for AMD or an OEM to create a utility that would allow you to adjust the frequency of a single core while turning the others down.
Well, according to the folks over at Expreview, they have such a utility being readied for the Phenom launch. Here is a pic of the utility.


So that means you don't have to run all four cores at 3.2GHz, you can just run on ethat high and keep the others at their default or lower speed. Tis will increase single-threaded performance and keep power down. I guess this is what they were talking about in the article a few days ago.

I was oping they'd say they got a retail chip and it's 10-15% faster than the ES. Still good news for AMD enthusiasts though.

That is one possibility but it also allows for tweaking of each core so all the cores can be pushed to their max. IE no 2 cores run at the same temperature if all have the same multiplier and fsb. Now if 1 core is holding you back you can lower its multiplier and increase the other 3.

This may save a small percentage of quads which other wise would be tri core. IE a 2.2GHz Phenom running 3 cores at about 2.3GHz an the fourth at only 1.8~1.9.
Related resources
a b à CPUs
a b À AMD
November 1, 2007 2:56:57 PM

I think thats mainly meant to be an overclocking feature, all cores will probably be running at the same speed when leaving the factory, so to speak...
http://www.nordichardware.com/news,6995.html
November 1, 2007 3:06:16 PM

This is good to hear. Shows the Phenom is a true quad core die.
Anyway, it looks like all 4 cores can be better managed depending on the workload.
November 1, 2007 3:38:51 PM

Does the Wikipedia definition of a Troll have the follow quote as an example? LO:love: 
Quote:
Aw, fer cryin' out loud. Give it up. Stick a fork in AMD and just walk away. Some of you guys (mostly Baron) belong in the Flat Earth Society. Phenom is an embarrassment, a failure, a joke, and the most uncompetitive product since the Edsel... Wake up and smell the benchys!

a b à CPUs
November 1, 2007 3:44:46 PM

What benchmarks? I haven't seen a RELIABLE benchmark for Agena as of yet. Until THG, Ananatech, or a reputable site does benchmarks, everything is speculation about its performance at this point.
a b à CPUs
November 1, 2007 3:46:33 PM

Quote:
Aw, fer cryin' out loud. Give it up. Stick a fork in AMD and just walk away. Some of you guys (mostly Baron) belong in the Flat Earth Society. Phenom is an embarrassment, a failure, a joke, and the most uncompetitive product since the Edsel... Wake up and smell the benchys!


It's always nice to hear from the peanut gallery. :sarcastic:  The last thing this thread needs is to turn into an Intel-fanboy-flame-the-Baron-fest. :non:  Here's $.25, now go post in another forum.

November 1, 2007 4:16:47 PM

Quote:
That's my point, bright eyes. Benches talk and B.S. (as well as B.M.) walks. I don't give a good damn about power draw or clock for clock. What does Phenom/Barcelona do? Is there any sane person on this earth (and we know who I'm excluding) who thinks that Phenom is going to beat a Q9650 in the next 18 months?

Phenom has more to worry about from the Q6850 in the short term as the penry has to run dangerously high 1.45 volts to get 4GHz. Penry while can do this risks burning out in as short at a few months at 45nm. The Phenom has to hit dangerously high 1.5 volts to get 3GHz but at 65nm its a little less risky. The 65nm Q6850 also runs high voltage for 4GHz but not much higher than the penry. Point is the penry may be forced to run at only 3.8GHz for a reasonably long life and the 7% clock for clock advantage over the Q6850 want hold verses a OC'ed 4GHz Q6850.

Most information was taken from tomshardware review of the Q9650.
a b à CPUs
November 1, 2007 4:26:51 PM

DELETED

Dont use that word again.

Turpit
a b à CPUs
November 1, 2007 4:38:16 PM

I agree, chunky. That is an impressive feature.
November 1, 2007 4:41:07 PM

DELETED
November 1, 2007 4:43:51 PM

I cannot remember, but this guy really sounds like someone who has posted under another name. CaptRobertApril? Maybe it was someone else, I cannot remember.
November 1, 2007 5:05:31 PM

Quote:
Drysocks, mangez mois SVP.

Are you trying to say bit me in french? Cause you failed at that to. :heink: 
I'm going to have to agree with the members here, this is impressive, if ti comes to be true.
Till the product is released I'm staying neutral, not jumping on either bandwagon till I see some number, and features?
James
November 1, 2007 5:08:30 PM

I think the utility is an excellent idea, and should enable overclocking of the best core for single threaded applications as Baron states. Despite the initially unimpressive showing of AMD's Barcelona, one feature I found envious over Intel's current implementation was the individual clock/vcore adjustments put into practice. It's ingenious, and it just might save the day for AMD. I look forward to seeing the experiments carried out with this tool.
November 1, 2007 5:14:53 PM

Quote:
Hey, Chunk, so it's ok for Baron to spam this forum for years with his B.S. and it's not alright for someone to come in and say the Emperor has no clothes? Who died and left you King... or Emperor... or Baron?



Do you know me from somewhere? I don't spam. I post relevant CPU news. Phenom was right behind Yorkfield with an engineering sample in the CrySis tests so what are you talking about. And in that test, the Intel dual cores were just as fast.
November 1, 2007 5:17:09 PM

elbert said:
Phenom has more to worry about from the Q6850 in the short term as the penry has to run dangerously high 1.45 volts to get 4GHz. Penry while can do this risks burning out in as short at a few months at 45nm. The Phenom has to hit dangerously high 1.5 volts to get 3GHz but at 65nm its a little less risky. The 65nm Q6850 also runs high voltage for 4GHz but not much higher than the penry. Point is the penry may be forced to run at only 3.8GHz for a reasonably long life and the 7% clock for clock advantage over the Q6850 want hold verses a OC'ed 4GHz Q6850.

Most information was taken from tomshardware review of the Q9650.



Perhaps you should see the post called "Will the real Agena please stand up!" It clearly states that for the retail version 2.3GHz is achieved at .9V so 3GHz should be around 1.3V, which is not too bad, but then it may be lower by the time AMD releases it.
November 1, 2007 5:19:27 PM

I would consider the Crysis demo to be lose-lose for everyone involved. Can't wait for real data to come in about Phenom X4 (and especially how the new architecture affects the X2). Too bad it seems AMD won't be selling Barcelona-based dual cores for quite some time.
November 1, 2007 5:27:47 PM

weskurtz81 said:
I cannot remember, but this guy really sounds like someone who has posted under another name. CaptRobertApril? Maybe it was someone else, I cannot remember.


I remember the Captain and he was fairly likeable, not like this guy at all. Maybe you're thinking of Rob, who posted under a variety of names before disappearing.
November 1, 2007 5:29:16 PM

Quote:
April? Yeah, he was one of the guys Baron and his gang of Jabba The Trolls pissed off so much that they fled to ***. If you guys really cared about this forum you would have banned Baron before you lost guys like that. But no, Junior Investigator, I've never posted here under any name. I've just been reading to try to get some valid information out of the pile of steaming crap that Baron and his friends have buried this forum under.



Wow, another disgruntled Intel employee, maybe. I may be whatever but I never come into a post and call names and disparage companies. Besides, I don't have any friends here, just people who post. Could it be you're mad that AMD has effectively switched their tactics due to what the server OEMs wanted?

They didn't want HT3 for Barcelona, so the actual 10h FULL IMPLEMENTATION is in AgenaFX.

An employee of AMD posted on several sites that B2F was a full 10-15% faster than the BA rev of Barcelona and with the added "bandwidth headroom" Agena should add another 5% or so. I'd say that will give similar clocked Yorkfield a run for its money.

You can disagree, but please refrain from the name-calling. As a matter of fact......
November 1, 2007 5:32:27 PM

OK... either way, I want to see good benchmarks. I want to see 3d marks scores, see graphs of cpu usage, graphs of how fast they can get tasks done. Non of the stuff that I have seen here has really proven anything yet. At least to me. I want to hear true credible benchmarks. Now, nothing against AMD or Intel, I just care about benchmarks.

For the posts about setting each core to a different speed. That may be a nice idea, but I think it lacks the usability. Who would want to have certain cores OCed to high clocks and others at slower clocks? The only good I see in this is for running at idle, for power saving. This may have been how they were able to get the agiea to run at so low voltage? Maybe, I do not know.

For quadhardcore: DELETED If AMD dies then how will I be able to afford my next CPU!!! Seriously think before you write!
November 1, 2007 5:43:36 PM

sailer said:
I remember the Captain and he was fairly likeable, not like this guy at all. Maybe you're thinking of Rob, who posted under a variety of names before disappearing.


This guy is a ****, but kind of funny. I think he just needs to calm down a bit and stop the attacks. If he could post some useful info and stop the above, everything would be fine!!
November 1, 2007 5:52:09 PM

Quad, does it really matter if it will out perform the Intel offerings? No. What is important is that it will perform WITH them. Put down your thesaurus and stop posting inflammatory crap.
November 1, 2007 5:54:00 PM

To post 13 posts and already be hated by everyone in this thread. Congrats, you did what most cannot (or try not) to do. Not even Rob was this bad to begin with. :ouch: 
While it is true Baron is incredibly bias towards AMD, he hasn't really posted any BS in this thread as of yet and only posted relevent information, therfor he hasn't deserved your responses. The information he provided was both interesting and it'll be nice if it comes true.
So please do us all a favor and DELETED. Thank you and have a great day.
James
November 1, 2007 5:58:57 PM

james_8970 said:
To post 13 posts and already be hated by everyone in this thread. Congrats, you did what most cannot (or try not) to do. Not even Rob was this bad to begin with. :ouch: 
While it is true Baron is incredibly bias towards AMD, he hasn't really posted any BS in this thread as of yet and only posted relevent information, therfor he hasn't deserved your responses. The information he provided was both interesting and it'll be nice if it comes true.
So please do us all a favor and DELETED.. Thank you and have a great day.
James



I'm not biased at all. I just have a preferred vendor. Try working on a P4 2.8GHz with 512MB RAM and perhaps you'll understand why I don't buy Intel. People like quad had no problem with the CrapBurst that is slowing down offices all over the world so I have a problem with him.

Intel is just a company. They do some questionable things, but I don't hate them.
November 1, 2007 5:59:46 PM

Looking at that image again, wouldn't adjusting the clock rates of a CPU on the fly cause any kind of lag?
James
November 1, 2007 6:01:13 PM

Speedstep and CnQ do it all the time :)  Why would manual adjustments be any different :) 
November 1, 2007 6:05:17 PM

If you have a preferred vendor then your biased to them, thats just how it works, or under my books anyways.
My families main computer has a 1.6GHz socket A AMD processor on 256mb RAM it's just as painful. It's like the P4's, they were decent at the time, but we are in the future. Any old technology doesn't scale up to the new.
James
November 1, 2007 6:06:38 PM

Good call thanatos421.
James
November 1, 2007 6:52:36 PM

I am an AMD fan myself. I guess I just like to pull for the underdog. Even when they had the performance crown, they were still the underdog, they just had a better product. I also am an ATi fan. I have owned both Nvidia and ATi cards in the past and I was always impressed with the image quality of the ATi cards.

Now, however, the system I am building is an Intel system with an Nvidia card. Why you may ask? They offer the best performance at this time. It doesn't really pain me to buy a competitor to my favorites, but it would have been nice to stay "loyal". I do BELIEVE (I don't know) that Intel will still hold the crown, but Phenom won't be as far behind as everyone thinks. I hope :) 
November 1, 2007 7:15:25 PM

BaronMatrix said:
Perhaps you should see the post called "Will the real Agena please stand up!" It clearly states that for the retail version 2.3GHz is achieved at .9V so 3GHz should be around 1.3V, which is not too bad, but then it may be lower by the time AMD releases it.


I would disagree with you on 0.9V for 2.3Ghz. I guess its a power saving feature. The Q6600 can run with 1.1V at idle. Not only that, there are other production samples (B2 stepping?) that run at full 2.3Ghz under 1.2V. In addition, the 0.9V is not consistent with AMD"s own roadmap, which classify GP9600 as a 89W part.

Nontheless, I guess we'll see more as time goes on.
November 1, 2007 7:30:01 PM

BaronMatrix said:
...I never come into a post and call names and disparage companies...

...Intel is just a company. They do some questionable things...
Never, huh? :sarcastic: 
November 1, 2007 7:32:29 PM

sonoran said:
Never, huh? :sarcastic: 


Of course. What are you talking about? Baron is one of the most informative, most intelligent, and most respectable poster on this forum. :D  :D 

Once he told me that "I'm a dependent child who needs AMD to have bad yields", and continue to say that "its an accurate statement judging by my personality".
November 1, 2007 7:40:50 PM

yomamafor1 said:
Of course. What are you talking about? Baron is one of the most informative, most intelligent, and most respectable poster on this forum. :D  :D 


Excuse me, are we talking about the same person? :pt1cable:  Or have you been sampling some of the wares that the guy on the streetcorner sells? :pfff:  Ok, Baron comes up with some good information from time to time, but as to that stuff about the "most respectable poster on the forum", well, yeah, that must be some weird stuff you bought. :ouch:  :non:  :non: 
November 1, 2007 7:41:53 PM

I think at this point DAAMIT will have to do what the Sony did with the PS3 and take a loss for a while even if they do make a superior product. I'm not saying that the PS3 is the kats pajamas but I'm saying that if you consider all that you get in the box and the fact that less than 2% of them fail compared to the 30% of the xbox360 you'll see what I mean.

Anyway I think that ATI is gonna have to bite the bullet and take a risk about putting something out there that totally kills NVidia instead of worrying about how it looks. I also think that AMD was ontop for a while and now it's their turn to take a step aside. Between Intel and AMD one of them will always be 2nd place, I just think it's gonna go back and forth about every 5 years now.
November 1, 2007 7:52:48 PM

I have to ask, is that pronounced Tater Raider (Rater)?
November 1, 2007 7:58:19 PM

What's with the handbag slapping today?
Rubber rod.
November 1, 2007 7:58:19 PM

BaronMatrix said:
Yes folks, my prayers have been answered. And AMD has an answer to Intel's single threaded prowess, sort of. What's he babbling about you say? Well, a few months ago I thought that with the split planes and individual core voltage it would be possible for AMD or an OEM to create a utility that would allow you to adjust the frequency of a single core while turning the others down.
Well, according to the folks over at Expreview, they have such a utility being readied for the Phenom launch. Here is a pic of the utility.


So that means you don't have to run all four cores at 3.2GHz, you can just run on ethat high and keep the others at their default or lower speed. Tis will increase single-threaded performance and keep power down. I guess this is what they were talking about in the article a few days ago.

I was oping they'd say they got a retail chip and it's 10-15% faster than the ES. Still good news for AMD enthusiasts though.



So what happens with the ram, if the ram is shared between all all four cores and the speed of the cpus are different how does the ram cope with this? Isn't ram speed tied to the speed of the processor?
November 1, 2007 7:58:28 PM

T8RR8R said:
I think that ATI is gonna have to bite the bullet and take a risk about putting something out there that totally kills NVidia instead of worrying about how it looks.
At the risk of going off topic (like that hasn’t happened yet in this thread, lol), I must respectfully say that I don’t understand this statement. 'Totally killing' NVidia is their objective, not a risk.
November 1, 2007 7:59:24 PM

sailer said:
Excuse me, are we talking about the same person? :pt1cable:  Or have you been sampling some of the wares that the guy on the streetcorner sells? :pfff:  Ok, Baron comes up with some good information from time to time, but as to that stuff about the "most respectable poster on the forum", well, yeah, that must be some weird stuff you bought. :ouch:  :non:  :non: 


That was for sarcastic purposes...

I know I know... I'll work on my satire skill..... :sweat: 
November 1, 2007 8:02:50 PM

intelamduser said:
So what happens with the ram, if the ram is shared between all all four cores and the speed of the cpus are different how does the ram cope with this? Isn't ram speed tied to the speed of the processor?


That's the job of the shared L3 cache, so AMD can clock four cores independently, while having the same bus speed.
November 1, 2007 8:10:47 PM

@Spongebob: I believe he meant to say that they need to worry more about raw computing power and maybe lay off the "features" so much. ATi still holds the image quality crown, but not too many gamers care about that as much as smooth game play from high fps, which Nvidia does MUCH better at this time.
November 1, 2007 9:10:42 PM

Very very interesting. looks like people will be able to enable a fourth core on a normally 3 core processor. Great move by AMD!
November 1, 2007 9:22:40 PM

Did you read the first post? :sleep: 
James
November 1, 2007 9:50:45 PM

james_8970 said:
If you have a preferred vendor then your biased to them, thats just how it works, or under my books anyways.
My families main computer has a 1.6GHz socket A AMD processor on 256mb RAM it's just as painful. It's like the P4's, they were decent at the time, but we are in the future. Any old technology doesn't scale up to the new.
James



From dictionary.com

#

1. A preference or an inclination, especially one that inhibits impartial judgment.
2. An unfair act or policy stemming from prejudice.
3. A weight or irregularity in a ball that causes it to swerve, as in lawn bowling.
4. The tendency of such a ball to swerve.

I judge impartially. I have no prejudice. I like AMD. Period. I used to only buy Intel as I bought into the "not stable enough" FUD regarding AMD. Now I only buy AMD. Maybe one day I'll only buy Intel again. Maybe I'll buy both.
November 1, 2007 9:54:56 PM

yomamafor1 said:
I would disagree with you on 0.9V for 2.3Ghz. I guess its a power saving feature. The Q6600 can run with 1.1V at idle. Not only that, there are other production samples (B2 stepping?) that run at full 2.3Ghz under 1.2V. In addition, the 0.9V is not consistent with AMD"s own roadmap, which classify GP9600 as a 89W part.

Nontheless, I guess we'll see more as time goes on.



Not true. AMD doesn't use TDP anymore they use ACP and the ACP for around that clock speed is 75W. Remember this is a quad core chip and my 2.3GHz 4400+ 939 runs at 1.35V. Twice the cores running at a lower voltage is damn good even though I'm at 90nm.
November 1, 2007 9:56:42 PM

intelamduser said:
So what happens with the ram, if the ram is shared between all all four cores and the speed of the cpus are different how does the ram cope with this? Isn't ram speed tied to the speed of the processor?



The idea of this is to do it for single threaded apps. AMD has a new write scheme that collects data into a cache and writes in bursts, so it shouldn't matter.
November 1, 2007 10:09:28 PM

BaronMatrix said:
Not true. AMD doesn't use TDP anymore they use ACP and the ACP for around that clock speed is 75W. Remember this is a quad core chip and my 2.3GHz 4400+ 939 runs at 1.35V. Twice the cores running at a lower voltage is damn good even though I'm at 90nm.


Unfortunately, we only know TDP published from VR-Zone. And according to the following graph on THG, the estimated power consumption is pretty close to AMD's publicized TDP (not ACP)


Therefore, I'm pretty sure Phenom having a TDP of 89W is close to its real power consumption. I'm not even sure if major 3rd party review site will take ACP seriously, because at the moment, none of them do.
November 1, 2007 10:35:45 PM

Thanatos421 said:
I have to ask, is that pronounced Tater Raider (Rater)?


You are correct...blame my dog, Tater(T8R). Raider just came from the rhyme. So T8RR8R it is.
November 1, 2007 11:18:19 PM

yomamafor1 said:
That was for sarcastic purposes...

I know I know... I'll work on my satire skill..... :sweat: 



Maybe the quotes in your sig throw people off?
!