AMD Turion 64 benchmarked

[danrok]

See it here:
http://www.coolaler.com/ipb/index.php?showtopic=42810

[dsio]

*skeptical*

i dunno... i really dunno.
Its unconvincing.

[MONKEYMAN]

definitely a bit sus

[Pitch]

Something is a bit dodge with the voltage.

1V on the second bench?

[nodavilp]

http://theinquirer.net/?article=21613
http://www.xtremesystems.org/forums/...ad.php?t=55174

[danrok]

Originally Posted by nodavilp

http://theinquirer.net/?article=21613
http://www.xtremesystems.org/forums/...ad.php?t=55174

Trouble is that those relate to the same source as originally posted, so they don't really confirm anything.

Hopefully, AMD will have this chip at CeBIT and we will get some official info.

[danrok]

It would appear the same guy now has a 4200+ CPU.

http://www.digital-info.net/vbb/showthread.php?t=2635

[dsio]

Just to illustrate the journalistic integrity of the inquirer, they source their latest tech news from, "Some dude, some where, from some country, who speaks some language, who has a picture, thats half blacked out, that was posted on some forum."

DAMN, brb, im going to that forums to post that i have osama bin laden living in my basement. After all, a photo of osama in my basement is all the proof you need, its not photoshopped or anything. Honest.

From the people who brought us the Geforce 6900, and the Athlon GT. well done guys, have a pulitzer!

[danrok]

It goes without saying, don't believe everythong you read!

What's a pulitzer?

[Sideburns]

A Pulitzer is a prize for damn good journalism.

The inquirer is a much ignored site for me.

[Bring the Noise]

I'm still wondering why people are thinking the FX-55 will equal a Athlon 64 4200+

Just by looking at AMD's past PR listings let make some comparisons...

3500+ 2.2Ghz 512KB L2
3800+ 2.4Ghz 512KB L2 (300 Pr points for a 200Mhz speed bump)
4000+ 2.4Ghz 1MB L2 (200 PR points for doubling the L2)
4300+ Min for 2.6Ghz 1MB L2 (300 Points for 200Mhz speed bump)

Cross pointing
From a 3500+ to a 4000+ we see 500 PR points from a 200Mhz increase and doubling the L2 cache. So we should see a 500 PR point jump from the 3800 to the FX-55's clock locked version, cause its the same 200Mhz increase and double the L2 cache.

Now the fact that the new E revision contains an enhanced memory controller (one that can run 4 sticks of PC3200 at PC3200 -vs- reducing speed to PC2700) and the addition of SSE3.

I'm thinking the PR number 4300+ or 4400+ for the 2.6Ghz cpu with 1MB of L2. Let the 512KB version be the 4200+

[Rondog]

All I get is ©êºp¡Aµo¥Í¿ù»~¡I*Õ*Y±z¤£½Tª¾¦p� �ó¨Ï¥Î¬Y¶µ¥\¯�*¡A©ÎªÌ¤£ª¾¹D¬� �¦ó±z·|±o¨ì¦¹¿ù»~°T®§¡A½Ð¸Õµ� �¬Ý¬Ý»¡©úÀÉ®×¥H¨ú±o§ó¦hªº¸ê °T¡C. I cant get any legible writing.

[dsio]

http://www.anandtech.com/cpuchipsets...oc.aspx?i=2374

Turion released. And its not a new chip at all, its the same old athlon64, only now they are calling the athlon64 mobile in combination with its chipset / architecture a "technology"

So, Turion 64 is to Mobile Athlon 64 as Centrino is to Pentium M. (after the cattle exporters are through with it)

[Bring the Noise]

AMD Turion 64 = speed bin Socket 754's that can run at minimal voltage while producing very very little heat paired with efficient chipset, graphics, wireless, and the rest.

Now if Nvidia can produce a version of the Nforce 4 ultra at 90nm and back it with a 90nm 6200 or 6600 mobility (8 pipes in the thin and lite market should be more then enough).
IBM's Fishkill Fab does 300mm wafers and 90nm process size, and Nvidia goes through IBM for some of their fab production (IE chips).

Back a speed bin'd Socket 754 with a 90nm chipset and GPU (all using SSDOI) and you should be able to come up with a thin-n-lite notebook that would give you decent performance with good battery life... This is what AMD needs to smack Centrino around!

Another thing to remember is AMD's agreement with IBM about SSDOI or what ever they are calling the Strained Silicon On Isolator process they have come up with.

So if Nvidia was still going through IBM for chip production. They could possibly use the production techniques that AMD has refined with the K8's E4 steppings but use them at IBM's plant (again AMD-IBM agreement on production techniques and the Fab's use the same tools on the backend).

That could create a low voltage chipset and more importantly a low voltage GPU that would not generate 60 to 80 watts of heat and guzzle electricity like a beast.

Anyhow just a hope.
But I'm thinking that 2Ghz isn't going to be enough for cpu speeds if Intel is about to push up to 2.2 and higher for the PentiumM cpu's.

[dsio]

I think it will be ok. Dothan does not have a whole lot of headroom and intel know that, which is why they are planning to change their dual core base from dothan to the next generation of the dothan design soon after their launch. We know the Athlon64 has headroom to burn however they have not limited turion 64 to athlon64 cpus. I would not be surprised if we saw an athlon64 / Geode hybrid sometime soon. Geode core, on 90nm, or possible even 65nm, given how successful their last die shrink was, with 64 bit enabled, and an on chip memory controller, with possibly a much smaller cache than you would expect on a mobile chip. The reduced DDR latencey from the on die controller makes the A64 much less effected by cache changes. Less cache could mean better battery life which is what i care about a hell of a lot more than performance. Performance around the 2.4-2.8 range with good battery life is alot more important than flat out speed.

I seriously think thats how they can beat centrino. Be making a slower performing platform with twice the battry life. I think Geode definately needs to come out of the woodwork in a big way, even if it does inherit some A64 tech.

[GrahamGarside]

This set of benches shows how little the L2 affects performance in most situations with the A64

http://www.xbitlabs.com/articles/cpu...on-2600_5.html

[dolanenwindrift]

The Inquirer is just a techie version of the Enquirer - Attentiojn grabbing headlines backed by poor writing, little or no fact checking, and no real technical information even when the facts are in order.

[dsio]

Originally Posted by dolanenwindrift

The Inquirer is just a techie version of the Enquirer - Attentiojn grabbing headlines backed by poor writing, little or no fact checking, and no real technical information even when the facts are in order.

Here here.

[psy^]

Originally Posted by dsio

Just to illustrate the journalistic integrity of the inquirer, they source their latest tech news from, "Some dude, some where, from some country, who speaks some language, who has a picture, thats half blacked out, that was posted on some forum."

But this is the Internet!
Nobody lies on the Internet