Tom's review??? What about anandtech's???

[DAPUNISHER]

Originally posted by: THUGSROOK
face it guys ~ they all suck.

you wanna know how good something really is?
buy it and find out

Succinct and accurate as usual :beer: However, I do trust some of youz guyz for a great "in the trenchs" review.

 

[rogue1979]

Well for anyone primarily interested in gaming it gives the appropriate information.

 

[Crashman]

I find it necessary to read both sites for complete and accurate information.

 

[Keysplayr]

Originally posted by: Elcs

Originally posted by: BDSM
It's very hard to put together a set of benchmarks that will be considered to be "fair". Since the P4 and the A64 are such completely different architectures they will perform very differently in different apps.

Not putting anyones views down here but there will always be people complaining about how hardware review sites perform their benchmarks. One of the major problems is the huge differences in the processors themselves. Programmers have a number of instruction sets from either side which they can choose to code into programs. A Processor is bound to perform better with programs that take advantage or better advantage of their capabilities.

Its impossible to perfectly balance bias but more effort could be made by the forerunners and most respected sites around.

Change the bolded word above from "processors", to "people". If there was ever really a problem, (and there really isn't) this would be a more accurate statement.

 

[ntrights]

Originally posted by: 1Crashman
I find it necessary to read both sites for complete and accurate information.

I have to agree, it's always been like that. Nevertheless i too feel that HT is underestimated (HT was the primary reason for my upgrade to P4c) very few benches show the benifit of running multiple applications like i think many of us do (except when you are strictly gaming competitive or programming) Duvie's real world benches shows us the benifit of HT in clear text. Having said that i think that Anandtech's final words in the review are very accurate (covers the strenght of both cpu's)

 

[First]

I don't really see what you're talking about Duvie. We included business, gaming, encoding, 3D rendering, and workstation apps. I didn't see any "synthetic" benchmarks used whatsoever.

 

[JWade]

this all reminds me of a few years ago when there was big heated discussion on benchmarks too, and for the life of me i cant find the multiple discussions on it.

basically AMD was using benchmarks that used 3DNow! optimized for their processors so fo course their processors would do better and intel would suck, intel would use SSE optimized benchmarks so the amd chips would suck (back during the P2/early P3 days and K6-2 and early Slot A days). One of the key things i remember reading is that it was able to make a p233mmx way outperform a 500mhz athlon slotA by messing with the bioses and different components involved.

I do think that the benchmarks by Duvie are good. But i must ask, but i am sure i know the answer already, do you have similiar configured athlon64/athlonxp/opteron systems to compare with? I know you asked for input for people that do. But the thing is, unless they are configured the same, not really accurate though.

Look at the beginning fiasco with nvidia and the FX line of vid cards and directx9.

I did take a poll of people i work with and know. out of 68 people, only 2 people do heavy regular encoding/rendering work. One has a P4 2.4ghz, other has a athlonxp2800+, both are happy, but cant compare them really as components inside the machine are different. The P4 is a Dell the athlonxp is custom home built.

I do think this thread is informative, but i do also think it is bias to intel.

This is coming from a person who went from an athlon xp2200 to a P4 2.4 and i cannot notice a difference in the games i play and the photo editing, occasional encoding i and my wife do.

 

[mamisano]

Why are the gaming benchmarks run at 1024x768? That taxes the video card and not the CPU. Almost all the gaming tests are equal for all processors because the card is the limiting factor...

 

[ntrights]

Originally posted by: mamisano
Why are the gaming benchmarks run at 1024x768? That taxes the video card and not the CPU. Almost all the gaming tests are equal for all processors because the card is the limiting factor...

Look at Open Gl and DirectX 7/8 games and DirectX9 Aquamark Cpu score they are not equal at all, FX51 and the 3400+ clearly show there strength in the gaming arena

 

[Cerb]

Originally posted by: mamisano
Why are the gaming benchmarks run at 1024x768? That taxes the video card and not the CPU. Almost all the gaming tests are equal for all processors because the card is the limiting factor...

With a GF2, maybe. 1024x768, no AA/AF is mostly CPU/RAM dependent now, using almost anything above a GF3.
As far as the benchmarks and such go:
1. HT is not underestimated. Look at the Business Winstone general use scores. Until the P4C, AMD dominated. While they still have an edge, HT brought Intel on par there. Older P4s did indeed feel much slower, which was reflected in those scores. With the 3.06 it became obvious HT was a very good selling point for P4s. They could now do what they started out sucking horribly at: multitasking.
2. Gaming is where a lot of differences show up, and alot of us are gamers.
3. Encoding matters very little. Face it: either you know what you're doing and can build a PC around it, or it's already fast enough. Most people are happy with going to bed after they start that DVD rip. Sure,we can get above real time, but until we hit 10x real time, extra performance won't matter that much to most people.
4. Having built and worked on many PCs, a lot of you here have, I can say with relative certainty that any PC of a given price is going to perform based on its parts more than just which CPU is in it. RAM, HD, mobo, video card and installed background apps make the difference, unless there is such a difference as 20+% between them. With 10% or less, the budget is the guide, and not much else.

Duvie's HT benchmarks are nice, and as I said previously, I'd like to see lighter multitasking benchmarks, running the normal benches with typical apps in the background, like IM clients and AV software.

 

[Tom]

"Most people are happy with going to bed after they start that DVD rip. "

I suppose that's true.. if you don't have HT !

 

[Jeff7181]

Originally posted by: Dead Parrot Sketch
"Most people are happy with going to bed after they start that DVD rip. "

I suppose that's true.. if you don't have HT !

As I said once before... usually people are either working with this type of stuff every day... or they're lucky if they work with it once a month. There really is no in between. I can see where it would be extremely beneficial for people who work with it every day, but for people who encode a movie maybe once a month, if that often, an overnight encode is no big deal.

 

[DAPUNISHER]

Originally posted by: Jeff7181

Originally posted by: Dead Parrot Sketch
"Most people are happy with going to bed after they start that DVD rip. "

I suppose that's true.. if you don't have HT !

As I said once before... usually people are either working with this type of stuff every day... or they're lucky if they work with it once a month. There really is no in between. I can see where it would be extremely beneficial for people who work with it every day, but for people who encode a movie maybe once a month, if that often, an overnight encode is no big deal.

I don't do DIVX or SVCD anymore and with DVDShrink, burning the backup takes as long or longer than the transcode. No photoshop, rarely MP3 encoding, and I have multiple systems that I can designate for different tasks so the lack of multitasking doesn't leave me waiting for 1 thing to finish so I can do another the way it does many who don't have a P4C. The only thing that had my hand waving back and forth over "add to cart" on a 2.6c was that I do SETI and HT would be sweet for me there. I've been gaming quite a bit lately though, and I already have a pretty fast card so I wanted to pick up frame rates for now via the CPU/platform upgrade and this 3000+@2.3ghz so far is very promising compared to my Barton@2.43ghz. Lot's more testing and gaming to do but the early conclusion is that I'm happy with it. Besides, I've been wanting a hammer generation CPU for 18+months now so it's also wish fulfillment BTW, the snappier feel of the Barton compared to my t-bred pales in comparison to the difference between the 3000+ and Barton, I guess thanks to the improved disk performance this AMD8151 NB ALI M1563 SB has over my nF2 and the on-die memory controller. Sorry for hi-jacking this thread I'll post up in the right place from here on out

 

[mechBgon]

Originally posted by: THUGSROOK
face it guys ~ they all suck.

you wanna know how good something really is?
buy it and find out

Hehe Very good approach, if you have the disposable income for it.

Duvie, remember that Anand has already covered the Hammer architecture pretty extensively several times. I don't think he wants to repeat it all again in every CPU review, but the "techie" stuff seems to be there. He just got it out of the way early on. Regarding the choice of benchmarks, I'm sure Anand is open to suggestion. Have you emailed him? Seriously, tell him how you feel. The squeaky wheel gets the worm, you know.

Here's one helping of food for thought, regarding benchmarks: think about how fast encoding apps change, and how many there are, and how frequently new builds come out. I bet you could name more than ten encoding apps, or variants of them, that you've used in the last year. Name one that you want Anand to stick with for as long as he stuck with Quake 3 Arena, or even just UT2003 Demo. If he does adopt it, how long will it be before you're unhappy again because you want him to switch to the hot new version with SSE3 support or 64-bit optomizations?

~ Throw them old results out, we're starting over again... again.

And now for a change of topic. I dare and challenge all i875P owners to lay out a concise how-to on getting an i875P board to run 2-2-2-5 timings stable at DCDDR400 or higher in the real world. What's the secret? Give your motherboard, RAM, voltages and settings please. I've had to advise many, many people to relax their timings and bump their DIMM voltage to stabilize their boards, and sometimes they turned hostile and INSISTED that the stuff MUST run at 2-2-2-5 or they will RMA it all for a second or third time, because it was SO important that it run at 2-2-2-5 timings.

~ whatever, dude... I'm gonna go play Mech4 Mercs for a couple of weeks while you work on that...