Have you all seen these Turion Benchmarks? - Page 2

Because Turion has an improved memory controller that allows it to run with full 200Mhz RAM speed and 1000Mhz HT (5 x 200, chipset willing), up from the 166Mhz RAM speeds & 800Mhz HT in the previous Mobile A64's.

Wow... this is a Centrino Dothan and it is killing it in almost every test. It seems to me that even in the float point physics calculations that the AMD will win but it won't kill the Sonoma set in the least. I think that it will hold its own. The only reason to buy a Turion when it comes out might be just for the price or maybe to hold out for XP64 which even then looks like a let-down.

I will wait until the Turions are officially released, but I might end up with a current Sonoma if it performs like this again.

BTW Both were running 6800 Ultra's... which Dothan has that?

Sweet! I always knew Intel had it in them

"the new standard"? Wtf? Are you joking?

Seems the author of this test lacks in knowledge, as he has no idea about chipsets already beeing available for Turion - just look at the last page, where he describes VIA K8T800 Pro as the "most "modern"" chipset available...

my take is.....

i am still waiting for turion rather than grabbing a Sonoma right now. Hopefully by that time hardware manufacturers will dish out some 64bit drivers

I dont think the new standard of operating systems will be 64bit just yet..

i would feel a waste to upgrade right now i just bought a copy of windows 32 bit two years ago... and i know a lot of people that aren't even planning to upgrade to xp yet

32bit will be around for a while i think..

my take is this:

I buy AMD for the same reason I use GNU/Linux and why I do not shop at Wal-Mart: I dislike the business practices of the market leader. I am willing to "suffer" some lower performance to support the company I like, just like I am willing to "suffer" missing some applications on GNU/Linux.

Of course, I would like to see some more benchmarks before forming an opinion about the Turion versus P-M.

always the against the current here..

everyone buy dell..i shall not
everyone buy intel.. i shall not (hold on
always the underdog for me..

Why do people feel the need to be radical and against the norm? The norm is typically the norm for a reason... it's you and you emo punks that look ridiculous when you go against the 'current.' If it has great performance and is cheap... then it is obviously a better deal than the same thing at an increased price.

"you will be able to run 64 bit applications and games soon (64 bit versions of Unreal Tournament, Far Cry and the Chronicles of Riddick are almost available; Stalker 2 and Battlefield 2 launches are not too far off)"

http://www.tomshardware.com/business...c_2005-01.html

How well does the Intel P-M perform in Windows 64 bit...?

If you guys think 64 bit will make a lot of difference, then you probably need a little theory on CPU's. 64 bit will allow for more memory, otherwise it really doesn't change a lot.

It's mostly marketing stuff.

Still, there's more registers on the new AMD's, that will help, there's also the fact that we can compute big numbers faster, and that's all there is to it, really.

And in my opinion, this will not change a lot... not in current software and games at least...

Now if you go play with a database, okay, that will do it, otherwise 32 bit is quite enough, more than enough I would say :P

But think what you want, I guess we computer engineers doesn't know our stuff.

the turion SUCKS in games, damn just look at the halo benchmark, oh well i'll stick to my pentium M even though im an AMD fan,

first i had an intel-powered dell with WIndows. (now a linux box...vida linux 1.1)
Now I have an AMD-powered Uniwill with Linux.

There was the ipod, and i have an iriver H320


market leaders are market leaders for a reason....the companies behind the product know how to advertise it.

Whoa those benchmark numbers were creepy. I was waiting for real benchmarks to see the turion blow the pm away. But jeeze, whats wrong with the turion? Time to rethink getting a PM instead

• Fact No 1--
  The Turion used in the test is the slowest of its breed, and one that is one speedgrade higher; OCed chips never perform as well as the chips that are actually designed to operate at that speed (more like tested in the fab process), it certainly does not do justice to the higher clocked Turions
  
  And they almost certainly do not have the optimized drivers at this stage with the chipsets for Turion.
  
  
  PHP Code:

  AMD Turion64 / Athlon64 System Configurations Processor(s) AMD Turion64 MT-34 (1.8 GHz) - 128k L1, 1024k L2 Cache AMD Turion64 MT-30 (1.6 GHz) - 128k L1, 1024k L2 Cache AMD Athlon64 2800+ (1.8 GHz) - 128k L1, 512k L2 Cache AMD Athlon64 3000+ (1.8 GHz) - 128k L2, 1024k L2 Cache AMD Athlon64 3200+ (2.0 GHz) - 128k L2, 1024k L2 Cache AMD Athlon64 3400+ (2.2 GHz) - 128k L2, 1024k L2 Cache
• Fact No 2--
  The chipset used in the Turion setup is designed for desktop NF3 250, and certainly is not the kind of chipset that will come out. It will certainly affect things such as tight control on thermal throttling, which are crucial in gaming. With a newer chipset, the Turions would almost certainly do better. The reasons why the Turion chips run better in Synthetic bench marks here is probably because the short length of the most synthetic benches. When longer tests are conducted, thermal envelopes are pushed to the limit, and the chipset would come prominantly into the forefront.
  
  
  PHP Code:

  Memory 2 x Samsung DDR-400 (PC-3200) Memory - 1 GB Total CAS 3,3,3 Latency at DDR-400 Speeds Motherboard DFI LANParty UT NF3250GB Motherboard nVidia nForce3 250GB Chipset nVidia Forceware 6.53 Chipset Drivers Installed
• Fact No 3-- From the descriptions of the gaming benches:
  
  
  Quote:

  Originally Posted by GamePC Unreal Tournament 2004 is a fast-paced DirectX 8.0 based first person shooter. The test runs through a series of intensive scenes with twelve AI-controlled bots. Average FPS scores are taken from the most CPU intensive scenario, Rankin.

  Quote:

  Originally Posted by GamePC Valve's Half-Life 2 is one of the most popular and system intensive shooters ever. We test using an in-house recorded fast-paced demo from the “Canals” area. The test stresses water and smoke shaders along with large outdoor environments.

  
  
  Quote:

  Originally Posted by GamePC Far Cry is a graphics-intensive DirectX 9.0-based first person shooter. Our benchmark consists if running pre-defined points through the game. The average frames per second for the runthrough is recorded.

  There is certainly a lot of room for subjectivity in selecting the type of data tha makes up the result of the benchmarks, as in the underlined sections.
  
  The precondition in the FarCry benchmark basically kills any advantage that Turion might have in handling the physics engiene, as the vertex data is already in the geometry storage of the system memory; I don' know why they even bother doing the test if they do that.

Therefore, I humbly submit the opinion that this review is skewed, just as 5 months ago, GamdPC did the same type of review that praised the OCed PM 2.4 beating the A64 3800+, which turned out to contradict every other review in the area.

I've seen how poor PMs can be at calculating a large number of moving objects with a sophisticated physics engiene, and it's not pretty, not in games like Children of the Nile, or in the AOE III that will be out in months. Perhaps PM can get ahead in shooter games by having optimized chipset features and large cache on first person shooters because the CPU in that case is not paramount anyways (graphic is), but the PM has proven time and time again to have inferior FPUs and small Branch Prediction Buffers.

I must say that I'm disappointed for a few reasons. One is that AMD seems to have decided to abandon their typical concern for the performance rating system. When the 3400+ is being outperformed by a different line of 3400+ processors, there's something wrong. Another is that Turion turned out to be, really, a pretty lousy extension of the Mobile Athlon 64 line. It's just not a competitor, with losses in battery life (compared to the Pentium M) and performance (compared to both.)

Now, though, there were some odd things about the benchmark. For example, I've seen some "interesting" benchmark results, but I've seen benchmarks where the 2.0 GHz Pentium M was barely outperforming the 3200+ (s939.) I must confess that I'm a little confused by the fact that the 1.6 GHz Pentium M beat the 3400+ on most of the gaming tests. Not only that, it simply raped it in the Halo test. I must confess to a certain degree of confusion, consternation, and suspicion.

For that matter, why did they measure the full system power consumption? Shouldn't they be testing the CPUs themselves? How are we getting a THIRTY WATT power difference between the Turion 64 and the Sonoma at full load? Unless they're lying about Turion thermal profiles (which is a possibility), the Sonomas must be providing power to yield advantages that great. The other possibility, assuming that all of the results are legitimate, is that the motherboards have enough power consumption to account for that. I'm not sure of how reasonable this is.

I already gave in and bought a Pentium M system (which may or may not be on its way to my home and due in half an hour.) I'm just voicing my suspicions. I don't care enough to do a full analysis and comparison to previous benchmarks, particularly since the benchmarks all have different goals.

I'd like to see benchmarks from another review personally before I cast judgement. I notice the Pentium M beats up regular A64 in their tests as well.. and that just doesn't seem possible to me.. since suposedly A64's are better than P4's which are better than PM's. But what do I know.

The answer to your first question is simply no. I have owned more computers with Intel chips than AMD chips in my life. I simply needed to point out the skewed nature of some of the tests, especially the gaming benches.

It makes me very suspicious that the A64 3400+ not just the Turion is beaten by so much in a test such as FarCry, which contrary to every other benchmark we have seen to date. A64 3400+ has topped PM on majority of the reviews in the past, certainly on games. Sure, PM 2.1 has barely topped the 3400 in some benches, but this would put the PM at nearly 130% of the performance of the higher clocked 3400+, which only lends to the suspicion.

After looking at the description in the fine print, everything becomes lucid. Of course; the way they are testing it, a P-III with superior graphics could probably beat an A64 on this form of test, if you are moving an object through pre-defined points in 3D space, there is no physics calculation at all. Of course then the PM would win with a bigger cache to access the geometry buffer faster! It's an oxymoron.

Unless fairer and much more rigorous tests are conducted, there won't even be close to a good comparison between the Turion and the PM.