AMD Releases 4 New 64s

[Bluehail]

http://www.techreport.com/reviews/2004q2/athlon64-3800/sandra-bw.gif
Wow

Full line of benchmarks at http://www.techreport.com/reviews/2004q2/athlon64-3800/index.x?pg=1

[Strawbs]

Originally posted by Bluehail

Wow
...nothing to add! except maybe ... :eek:

[iceblue]

There are Opterons but no Xeons.

[megaspazz]

amazing benchies... :D

[Logan2002]

AMD has the intel cpus SMOKED!

Ouch!

[Zappattazz]

It's always interesting when a "unbiased" site does a "comparison" between AMD and Intel chips. They always say "we did our best to deliver clean benchmark numbers". Usually, one only has to review the test system configuration to see how they achieved their "best".

http://www.techreport.com/reviews/2004q2/athlon64-3800/index.x?pg=3

They use top-of-the-line ASUS/MSI motherboard with the AMD setup, but choose a middle-of-the-road ABIT IC7-G for the Intel chips. Why not use the best Gigabyte/ASUS models?

http://www.tomshardware.com/motherboard/20030707/i875p-71.html

They used the fastest memory timings with AMD tests, but for the Intel tests they used less the optimum settings

http://www.tomshardware.com/motherboard/20040119/index-12.html

We observed one interesting result in many of the gaming benchmarks: while the Pentium 4 3.2 GHz is normally just a touch faster than the Athlon 64 3200+, it quickly falls behind the Athlon if you only use slow memory modules.

[doubleclick]

lol, there's some truth about what Zappattazz says...

Nice CPU's nonetheless...

[Strawbs]

As if to prove that when taken out of context, you can make any review say anything you want it to.

The following Quote is from the same link "Z" posted: http://www.tomshardware.com/motherboard/20040119/index-12.htmlthe Pentium 4 processors either take the lead or bring up the rear, depending on whether the memory timings are fast or slow. We were duly impressed by the Athlon 64 FX-51's scores, which maintained its ranking no matter what kind of memory it was given. This steadfastness is largely due to the integrated memory controller.

One might add that the fastest available 64bit Athlon boards are probably the nForce3-250's (http://www.anandtech.com/chipsets/showdoc.html?i=2004&p=1), of which, precisely none were used.

[porsch1909]

one benchmark oh dear. i encourgage everybody to see all the benchmarks. and youll all see that intel win some lose some. and the ones intel lose are by tiny margins and the ones intel win are by tiny margins. its a joke!!

why do they incluse the prescott or E intel CPU its rubiish i hate it, the C is better. the only thing the prescott is good for is rendering and encodeing etc.

this is AMDs third wave of 64 bit CPUs and intel still have to bring their first. what are they playing at, probably going to skip 64 bit and bring 128 bit out first.

:D

[causticVapor]

The problem is more with the processors' pricing than the opinions and strategems of one little hardware website which is already looked down upon by most enthusiasts. $500 is way too expensive for a midrange chip. At this rate, AMD will price themselves out of the market like Motorola.

[Someone Stupid]

Hector "Chapter 11" Ruiz :)

[Zeekfu]

I'm sure it is all mostly hype anyway. It is imho basically impossible to compare apples to apples when putting the different cpus up against each other. The mobos, chipsets, and sometimes the memory are by definition not exactly the same.

It is interesting how I recently looked back at that Tom's article and interpreted as just the opposite. I noted in that memory timmings article they were able to use 4 sticks of memory with better timmings on the intel set up.
----------------------------------------------------------------------------------
While it's next to impossible to find one single RAM module that will work on all four platforms, that won't really affect our test results. We weren't aiming to compare different platforms, but to see what kind of an impact best-case and worst-case timings would have on performance.

We used the fastest-possible memory configuration for each platform. For the Pentium 4 and Pentium 4 Extreme Edition on the Asus P4C800-E (Intel 875P), we used four 256 MB DIMMs from Corsair because they allowed us to set the fastest-possible timings of CL2.0-2-2-5.

This was not possible with two 512 MB DIMMs because the fastest ones we could find - also from Corsair - supported "a mere" CL2.0-3-2-6.

That means that we had to test all the other platforms because the Athlon 64 FX and the Athlon 64 would only run stably with four DIMMs and much slower timings, while the nForce2 board from Asus only ships with three DIMM slots. For the Athlon 64 FX, we swapped out the Corsair RAM for a matched pair of Legacy Electronics modules.
-------------------------------------------------------------------------------
So don't get me wrong I really like Tom's hardware. I am a newbie and they give me a clue. I will say that the intention of the article was stated as: "We weren't aiming to compare different platforms, but to see what kind of an impact best-case and worst-case timings would have on performance." Now the first time I looked at this article I was simply amazed at the technology because I didn't have a clue what they were talking about. So once again I gotta give Pabst his props.:)
When I got closer to making the upgrade I had been planning on making for some months now I looked back over the article though. Despite the stated intention of the article the platforms were stacked up in the same graph and the intel platform had much better memory timmings, giving it the edge in my opinion.

I am not sure what you are referring to on that middle link. It is a roundup of all intel boards. If you are comparing benchmarks from that article to another it is hard to to because the test set up might not be the same. Or perhaps it is the wrong link or I misunderstand your point.
As reguarding the first link to the other site I don't know much about that site. I think sometimes the choice of mobo might be what is at hand or in other instances what has been recently tested using the same basic test platform. I imagine many of these sites would try to keep their test platforms tight at least in the short term so that every time they get a new toy in they don't have to run all those test over again. At any rate although I am sure it is done as a practical matter at least occasionally that introduces another variable. If test are not done in a controlled environment something like the tempature or humidity could possibly have some (albeit minor) influence on the test from day to day, hour to hour, week to week.
Anyway, I am not an expert by any means. I am getting ready to upgrade about a 7 year old system and looking for the best buy within reason for the money. I was aiming heavily towards p4 in 2.8-3.2c range even though I had gone amd on my two previous builds. This leaning towards p4 was primarily based on my impressions of Tom's hardware end of the year round up an so forth. When I started asking questions people had a lot to say. I have looked into athlon 64 more and think that may be the way to go for me. Still thinking on it.
Anyway, I am just making the point that I had a totally different opinion than you based on basically the same information. I looked back over the Tom's stuff and said to myself, "gee maybe they are weighted a bit towards intel now" and "I shouldn't close out the idea of going athlon 64".
Isn't human nature grand? :P
Abit boards have been good boards for me in my very limited experience. I don't know about that specific one you mentioned though.

[Vampiel]

Originally posted by Zappattazz
They used the fastest memory timings with AMD tests, but for the Intel tests they used less the optimum settings

No they didnt...
http://www20.tomshardware.com/motherboard/20040119/index-03.html

Intel
DDR400 (200 MHz) Corsair TWINX PC3500
2x 512 MB CL 2.0-3-2-6 (Socket 754 and Socket A)
4x 256 MB PC3500, CL 2.5-2-2-5 (Socket 478)

AMD
Legacy Electronics Registered ECC PC3200
2 x 512 MB CL 2.0-3-2-6
(88S6HDAR-1TDG 5733) Infineon Chips

Looks like Intel had the clear advantage, b/c the FX needed Registered memory, that's why they had to do that.

[Zeekfu]

After reading your comments and rereading the original post I just realized why the middle link went to the all p4 board article, duh. Ok I see the abit board is sort of "middle of the road" as he says on the benchmarks. However, on most all of those benchmarks those chipsets are fairly neck in neck regardless of where the board shows up in the line up. The difference between the top board and the middle board is ususally very minor % wise
Now that I understand zapp's origional posting better I would have to say that I disagree with the statement about which boards he thinks are better. Tom's gave the gigabite board a thumbs up based more on features than performance. As it looked fairly "middle of the road" in many of the bechmarks as well as far as I could see, albiet a little higher up than the abit boards in general. The msi was a clear performance "winner" which was suprising considering it uses the "slower?/cheaper?" 865 chipset instead of the 875 chipset. The two boards were recognized for not only their performance but for their price/performance/features presentation overall (at least in the case of the gigabite board)
And in the case of the memory timming article the stated purpose was to compare the memory timming at best-average-worst within their respective platforms but since it all would up on the same benchmark sheet I would submit they should have used two 512 dimms for the p4s with the same timmings as the rest.
Yea they had to use registered for the fxs but not for the regular athlon 64s
----------------------------------------------
While it's next to impossible to find one single RAM module that will work on all four platforms, that won't really affect our test results. We weren't aiming to compare different platforms, but to see what kind of an impact best-case and worst-case timings would have on performance.

We used the fastest-possible memory configuration for each platform. For the Pentium 4 and Pentium 4 Extreme Edition on the Asus P4C800-E (Intel 875P), we used four 256 MB DIMMs from Corsair because they allowed us to set the fastest-possible timings of CL2.0-2-2-5.

This was not possible with two 512 MB DIMMs because the fastest ones we could find - also from Corsair - supported "a mere" CL2.0-3-2-6.

That means that we had to test all the other platforms because the Athlon 64 FX and the Athlon 64 would only run stably with four DIMMs and much slower timings, while the nForce2 board from Asus only ships with three DIMM slots. For the Athlon 64 FX, we swapped out the Corsair RAM for a matched pair of Legacy Electronics modules.
----------------------------------------------------------------------------
translation:
-In the case of the athlon 64s the chips ran at faster timming with the two modules rather than 4.
-The test board they used for amd xp cpu only had 3 slots so no way to use 4 memory modules
-athlon 64 fx uses registered dimms (which by the way are by a completely different manufacturer but picky picky.)

All in all I think Tom's is a fairly accurate and objective site. I have learned alot from reading over their reviews. Alot of it I find is over my head, at least at first, but then I look into it and learn a bit more (hopefully).

In summary the stated purpose of the article was not to compare the platforms but to compare the effects of memory timming within the platforms.
the "con":
perhaps most(many?) of us just see a benchmark (any benchmark-not referring to Tom's excellent site here and get mesmerized but the colors of the bars without really looking at what the test is all about and % wise are the results significant/within a margin of error/flawed by testing set up/etc.

anyway I am just a nooobie myself ...just postin' a thread :p
peace

[Vampiel]

Toms is a very reliable site, they are not biased, ive built computers for 11 years and they always seem to know what they are talking about. Ive noticed some people that ARE biased do not like tom's so that just tells you something in itself right there.

[Vampiel]

Originally posted by Strawbs
As if to prove that when taken out of context, you can make any review say anything you want it to.
Yep

[Zappattazz]

I agree that Toms Hardware is unbiased - if you want the straight dope, they are "the site".

At the beginning of this thread, Bluehail posted SiSandra memory "timings" {see the spelling Zeelfu} that tauts the AMD chips performance.

My comments:

They used the fastest memory timings with AMD tests, but for the Intel tests they used less the optimum settings

are referring to Bluehail's link for

http://www.techreport.com/reviews/2004q1/athlon64-fx53/index.x?pg=1

It is at this site where they used substandard mobo/memory timings to bias the performance of the AMD chips

[doctj]

I think the 939 cpus are overpriced. A fine move if you want increased market adoption:rolleyes: .