K8L is actually K10..... apparantly

[Julian33]

Please don't kill me for linking from the Inquirer!

http://www.theinquirer.net/default.aspx?article=37444

The only reason I post this is because it does have more credibility than normal Inq stories because the details come from Giuseppe Amato who is actually part of AMD.

If they have gone with K10 then it does suggest that the Barcelona chip and it's derivatives are a total overhaul as opposed to a few minor tweaks here and there. I guess it's also possible that we'll see a new name instead of Athlon 64 - they have changed name at every major revision AFAIK.

 

[piesquared]

Please don't kill me from linking from the Inquirer!

http://www.theinquirer.net/default.aspx?article=37444

The only reason I post this is because it does have more credibility than normal Inq stories because the details come from Giuseppe Amato who is actually part of AMD.

If they have gone with K10 then it does suggest that the Barcelona chip and it's derivatives are a total overhaul as opposed to a few minor tweaks here and there. I guess it's also possible that we'll see a new name instead of Athlon 64 - they have changed name at every major revision AFAIK.

Once numbers come out, I imagine there will be a few more people adopting the AM2 platform for the option to upgrade when dual core Barcelona's come along (just a guess).

 

[piesquared]

This might also interest some. Nice little interview on hexus tv.

http://www.hexus.tv/show.php?show=28

 

[1Tanker]

This might also interest some. Nice little interview on hexus tv.

I]http://www.hexus.tv/show.php?show=28

I like the way he keeps mentioning best-to-market is better than first-to-market(which they've continuosly bragged about over the last few years). They've really changed their tune in the last ~9-10 months.

 

[Julian33]

Haha, it's pretty ironic that the ad at the beginning of the vid is for the Core 2 Extreme procs :lol:

 

[DavidC1]

No, the newer processors aren't KX's anymore. It's just like how Intel doesn't brand their CPUs, P6/P7/etc. According to the article, the dual core Athlon, the X2 is K9. X2 doesn't really change anything architecture-wise.

From Charlie, who originally found the name, the K8L name was made up by Intel.

 

[Bluefinger]

Well, to be fair, AMD had to change their tune considering what Intel is doing with C2D. Though from what I've heard in the interview... Barcelona sounds mighty tasty... I just hope its as tasty on release as this guy says it is...

 

[Pippero]

Please don't kill me for linking from the Inquirer!

http://www.theinquirer.net/default.aspx?article=37444

The only reason I post this is because it does have more credibility than normal Inq stories because the details come from Giuseppe Amato who is actually part of AMD.

If they have gone with K10 then it does suggest that the Barcelona chip and it's derivatives are a total overhaul as opposed to a few minor tweaks here and there. I guess it's also possible that we'll see a new name instead of Athlon 64 - they have changed name at every major revision AFAIK.

Who said "a few minor tweaks"?
IMO, K8L or whatever you want to call it, presents more uArch changes from K8, than K8 did from K7.
However, still IMO, the changes from K6 to K7 (or Pentium 3 to Pentiun 4) were waaaay bigger (that's what i would call a completely new architecture).
Concerning the "K10" code name... AMD did a lot of strange stuff.
First K8, it was supposed to be a totally new core, 6 issue 8O with extremely advanced HW scheduling and probably low clock (this was suggested from a series of patents which never made it into any silicon).
Instead, they scrapped the project and beefed up the K7 with AMD64, HT and IMC, and called it K8.
Then K9, which was cancelled altogether (rumored to be 8 issue, according to wikipedia). (link)
K10 was supposed to scale to 10GHz! (link)
K8L has a lot of advancements compared to K8 and will be signifcantly better in terms of IPC, but what's a bit disappointing is that it doesn't seem to bring anything really innovative to the table.

 

[piesquared]

I wonder if SiGe was implimented on rev F3 as in the 90nm 6000+. Doing benches at 3.5 ghz on air is pretty good. We'll see if it's true though.

http://www.hardforum.com/showpost.php?p=1030595627&postcount=2

 

[DavidC1]

K8L has a lot of advancements compared to K8 and will be signifcantly better in terms of IPC, but what's a bit disappointing is that it doesn't seem to bring anything really innovative to the table.

Yes. The thing I want to tell people is bringing the chip design to production takes several years. Unless AMD had information about Conroe as soon as Intel started developing the design, K8L isn't AMD's answer to Conroe. It's just their own advances. The only thing AMD can really do against Conroe is have sufficient clock speeds. On the other camp, Intel didn't put out the Merom core to compete against Athlon 64. The specific chip that was made to compete against Athlon 64 was the Conroe. The Merom design has been in development for number of years. The difference between Conroe/Merom is again differences in clock speeds and FSB. Major design changes aren't possible in a flash.

Conroe/Barcelona(stop using K8L, Intel stopped using Px naming and AMD stopped using Kx naming, and the talk is K8L is a name made up by Intel) both brought their own enhancements to the table. It's just that Barcelona has been designed later(possibly) than Conroe.

 

[Pippero]

On the other camp, Intel didn't put out the Merom core to compete against Athlon 64. The specific chip that was made to compete against Athlon 64 was the Conroe. The Merom design has been in development for number of years. The difference between Conroe/Merom is again differences in clock speeds and FSB. Major design changes aren't possible in a flash.

I don't understand you.
Conroe / Merom are exactly the same architecture.
As you said, the difference is just FSB and clock speed (and power consumption).

 

[DavidC1]

I don't understand you.
Conroe / Merom are exactly the same architecture.
As you said, the difference is just FSB and clock speed (and power consumption)

You probably want to read the quote I put on my response. Merom isn't the response to Athlon 64, because of the development time. Any additional improvements to make it competitive is made up by factors that doesn't need 3+ years to change, which is clock speeds and FSB speeds. Conroe is the answer to Athlon 64 because of that. Has Netburst been better, we would have seen Merom only for laptops.

 

[dasickninja]

Yes but, Merom = Conroe. And also Conroe = Allendale. Not to mention Conroe = Kentsfield. You see the point? Its different flavors of the same core.

 

[DavidC1]

Yes but, Merom = Conroe. And also Conroe = Allendale. Not to mention Conroe = Kentsfield. You see the point? Its different flavors of the same core.

I see your point, but you are not seeing my point. As I will say it again, go read the quoted paragraph.

Merom is not made to compete against Athlon 64, contrary to those people thinking it was. Merom was first seen on roadmaps on japanese/chinese sites since Pentium 4's were the rising stars. The actual development on the core happened even longer than that.

It's Conroe. I don't know how many times to say this but design to production takes 4-5 years. Clock speeds and FSB speeds tell how competitive the chip will be, and that's why Conroe is meant to compete against the Athlon 64 based cores.

Had Pentium 4 Prescott lived up to people's expectations, would we have seen Conroe?? Not likely. We would have seen it in laptops for sure though. But Prescott went bad, really bad, which is why Conroe came.

 

[dasickninja]

Yes I see your point. Netburst for the desktops and Merom for the laptops, Merom being an update of the Pentium M.

 

[DavidC1]

Yes. But because Netburst died, Conroe came.

 

[dasickninja]

I think they would have introduced Conroe earlier, possibly for the server sector.

 

[Julian33]

Well someone mentioned earlier tha the jump from K7 to K8 wasn't huge.... thats certainly true from what I remember. I'd describe K8 mainly as being a platform enhancement as opposed to a huge boost in core performance - I believe the early K8's didnt have much of a lead over equivalently clocked K7's. I would expect the changes in this chip to yield greater performance gains, so 40% does seem reasonable - again I do question how much of Core 2's performance gains over the K8 architecutre is down to massively beefed up SSE execution (which Barcelona and its derivatives will be getting).

The interesting thing is that with a new architecture, this presumably means the Athlon 64 name will go, as the name has changed with every previous increment I can remember in the past (guess thats my logo out the window ). As for the future, who knows? Will Shanghai be another big step forward and be K11, or will it just be a minor increment on K10 as the K8 has been enhanced over the years?

My bet is looking into the longer term, the next big overhaul will be when Fusion comes along. That could be such a radical change that AMD ditch the K system entirely. I think that what will probably happen is that an all new architecture built from the ground up will appear to support Fusion.

 

[DavidC1]

I believe the early K8's didnt have much of a lead over equivalently clocked K7's.

Actually it had quite a lot of performance boost over K7. The IMC alone brought majority of the performance improvements(~20%)
http://www.anandtech.com/showdoc.aspx?i=1884

My bet is looking into the longer term, the next big overhaul will be when Fusion comes along. That could be such a radical change that AMD ditch the K system entirely.

Take the K naming system from AMD and R naming system from ATI and make KR processors. Hell, even name it KilleR .

 

[BaronMatrix]

Please don't kill me for linking from the Inquirer!

http://www.theinquirer.net/default.aspx?article=37444

The only reason I post this is because it does have more credibility than normal Inq stories because the details come from Giuseppe Amato who is actually part of AMD.

If they have gone with K10 then it does suggest that the Barcelona chip and it's derivatives are a total overhaul as opposed to a few minor tweaks here and there. I guess it's also possible that we'll see a new name instead of Athlon 64 - they have changed name at every major revision AFAIK.

AMD never officially named it K8L. That's why I used Barcelona even though it's more typing. They have said before that it is a radical redesign. It'll probably still be A64. Just higher model numbers.

 

[Julian33]

Ok, I stand corrected. I seem to remember seeing some benchies in the past that didn't show much difference in certain situations, obviously I was wrong.

I like that naming scheme btw :lol:

 

[BaronMatrix]

This might also interest some. Nice little interview on hexus tv.

I]http://www.hexus.tv/show.php?show=28

I like the way he keeps mentioning best-to-market is better than first-to-market(which they've continuosly bragged about over the last few years). They've really changed their tune in the last ~9-10 months.

He only meant in terms of quad core.

 

[Bluefinger]

I wonder if SiGe was implimented on rev F3 as in the 90nm 6000+. Doing benches at 3.5 ghz on air is pretty good. We'll see if it's true though.

http://www.hardforum.com/showpost.php?p=1030595627&postcount=2

Interesting. I wonder if this means that Barcelona may have the potential to overclock like a beast. Speculation once more, but its the only thing one can do until AMD releases some engineering samples for benches.

 

[Pippero]

I'm not so sure about this analysis...
Intel has separate design teams, which are somehow in competition with each other.
To my knowledge they have 3 teams, the Itanium one, the Israelian one (Conroe), the Netbust one :lol: .
I think the Israelian team designed their latest uArch to serve multiple segments, and especially the server market is very demanding as it combines the power and thermal efficiency of laptops with the raw speed of desktops.
Also, Prescott debut was 3 years ago, and it's extremely likely that Intel was aware of the power wall this architecture was running into at least 6 months or a year earlier, this would have given enough time to the management to switch to the Israelian branch for the next all around core.
As a matter of fact, Merom doesn't seem to me as an architecture fully tailored for the mobile market, it's mora a jack-of-all-trades, in fact Core 1 Duo is better when it comes to power efficiency, and right now there exist ULV versions of Core 1 Duo which consume only 1W.

 

[tamalero]

Well someone mentioned earlier tha the jump from K7 to K8 wasn't huge.... thats certainly true from what I remember. I'd describe K8 mainly as being a platform enhancement as opposed to a huge boost in core performance - I believe the early K8's didnt have much of a lead over equivalently clocked K7's. I would expect the changes in this chip to yield greater performance gains, so 40% does seem reasonable - again I do question how much of Core 2's performance gains over the K8 architecutre is down to massively beefed up SSE execution (which Barcelona and its derivatives will be getting).

The interesting thing is that with a new architecture, this presumably means the Athlon 64 name will go, as the name has changed with every previous increment I can remember in the past (guess thats my logo out the window ). As for the future, who knows? Will Shanghai be another big step forward and be K11, or will it just be a minor increment on K10 as the K8 has been enhanced over the years?

My bet is looking into the longer term, the next big overhaul will be when Fusion comes along. That could be such a radical change that AMD ditch the K system entirely. I think that what will probably happen is that an all new architecture built from the ground up will appear to support Fusion.

wasnt there alink in vrtech saying that Shangai will be bring HT3 ( up to 5400 mts ) insteath of the usual 2000 mts in Barcelona, Barcelona AM2 models will have Ht3 too ( but not for socket F, 'til Shangai)