Duron 850 CPU Benchmarks

ravedaddy and quite a number of other folks wrote in with the news that Sharky's looks at the processor which benchmarks very well in comparison to Intel's 800MHz Celeron - the AMD Duron 850. Last week, with the release of the Celeron with a 100 Mhz FSB ^[?] , Intel jumped forward - while AMD's Duron has an equivalent 200 Mhz bus (100 Mhz buses). It looks like AMD is keeping the crown in the "Value" category.

Comments

Duron is a Great Chip

[cansecofan22]

I know that the Duron has been slow to take off because of a lack of an integrated chipset but I have been using a Duron 600 for a while now and with the 256 MB of RAM that I picked up for $100 it kicks ass. The chip is inexpensive (the 850 will be around $150) and the board with onboard sound costs about $100. Add 128 MB of RAM for $50 and a Voodoo 3 AGP card for about $70 and a $100 Maxtor (or Segate) 20 GB Hard drive and you have a fairly inexpensive system that will perform as well as any PIII 700 out there. And you are olnl talking a couple of hundred dollars. AMD is beating Intel all around. Now if they would only get there act together for a dual CPU system, there would be no holding them back!

Not just the value market

[Oscarfish]

A couple of weeks ago, I bought a T-Bird 1000 MHz chip and an Abit KT7-RAID. I got home, set it up, connected the L1 bridges, and stuck an Alpha PEP66 on it. I'm currently running at 1313 MHz (12.5 * 105).

The chip (the T-Bird) is a blue core, which means it came from FAB-30 in Dresden. From the pictures at Sharky's you can see they had a green core, which mean it came from Austin (and uses aluminum, not copper interconnects).

Re:Not just the value market

[gle]

The color comes from the finish, not from the technology used. So if it's blue, it does not much more than if if was yellow or pink or with a Nathalie Portaman picture on it.
Look at the letters and numbers on the core.

Re:Not just the value market

[Teferi]

Actually, the color -is- important.
A blue core means it's from the early line of Tbird/Durons, and it's possible to unlock the multiplier by connecting the bridges with something conductive.
Green-core chips are later, and can not be defeated in this way - I should know, I have one. :/

"If ignorance is bliss, may I never be happy.

Re:Competiton is what we need

[doctor_oktagon]

the more competition there is between chip makers, the sooner we will have faster more efficient machines, so I'm all for it

Unfortunately the history of computers has shown us that increased competition can also lead to overcrowding in a very saturated market, and the most interesting technology dissapears because the manufacturer can't market it properly (i.e. put the most spin on it).

Draw parallels with the home computer market of the late 80s for instance.

I think competition between AMD and Intel is great, but I'm not sure the market can handle any more big-scale players - hey ... look at Citrix: they have all but disappeared to the average PC consumer.

We need a better benchmark, and we need it soon!

[Flabdabb Hubbard]

I cannot be the only one who has trouble comparing processors these days. What with pipelines and prdictive branching, not to mention the various front and rear-side bus speeds, internal and external, it is harder and harder for me to look at the MHz rating to work out which is the best CPU.

Perhaps we should devise a new number, (call it the power rating) which would multiply all the numbers (bus speed, data path width, pipline parallelism, etc) together to come up with one easy to compare number that I can use when I buy my next PC.

Otherwise I have to know far more about computer architecture than I really want to. How about it CPU manufacturers ? A single number we can compare ?

AMD vs. Intel

[CoBoLwArRiOr]

I've always heard people telling me that Intel was better than AMD, and AMD has had compatibility issues with Windoze. I had an Intel 486. That was the first and last Intel I've ever had. I haven't had one problem with AMD in the compatibiliy issue. When it comes down to it, price and performance don't lie. I'll take my AMD anyday. Durons seem to be decent chips for the price, but for power and speed (aren't we all speed demons?) I'm going to go with an Athlon Thunderbird for my next machine. As for the 'Intel is better than AMD' or 'AMD is better than Intel' battles that wage on, I feel that the best chip is whichever one does the job for you. For me, that's AMD.

-=-=-=-=-=-=-
The COBOL Warrior

Re:We need a better benchmark, and we need it soon

[djocyko]

I very much disagree with you on thhis case. I have been developing a webpage that is dedicated to comparig setups via gaming benchmarks. Why, you ask? Well, when it comes down to it, those are the numbers that count. Real world numbers. Who cares how many megs the bus can push or whatever? It is the final outcome that matters. To this extent, web sites that curently use all these gaming and professional becnhmarks to test the setups are doing exactly what we need; they are giving us the numbers that count in the end.

I think the last thing we want to do is decide on some arbitrary index to rate all cpus. It would be a synthetic number that doesn't mean anything in the real world aside from bragging rights. It's like the Mhz war. It's a whole lot of bull. What matters is the final outcome, not the theoretical. So, my input is that we must stay away from that golden standardized index.

Sure, this makes it harder for the average consumer who goes to compusa planning to buy a computer, but at this point, anything you buy is damned good, so I am not horribly concerned about that.

But will AMD change the socket type?

[X-Dopple]

Right now, I love it. I can freely interchange my Duron and Thunderbird (not that I have one) on my Socket A motherboard. I don't have to put up with any socket type incompatibilities (Socket 370, Slot A, Socket 7) with my CPUs

But, down the road, will AMD change the socket type for its upcoming 64-bit Sledgehammer chip?

Re:Folks, you just don't get it ;-)

[toofast]

Yes, I use an 8-way SMP box boasting 8 Pentium III Xeon's @ 800MHz, with 2MB cache, and 1GB RAM as my home desktop machine. I just installed the Mylex RAID controller yesterday. The lights dim when I power it on.

My operating system of choice for this puppy: Windows 98. Beautiful SMP support, especially after I tweaked and recompiled the kernel. I get _at least_ 100fps in Quake3, in software rendering emulation mode.

Entire Bogusness.

[toofast]

If I can run Linux web/smb/inn servers on Athlons without hics -- if I can play Q3A, UT, SoF one after the other, on Athlon and Linux, and still work with my machine after that, the chip isn't the problem.

The buggy chip and incompatibility issues have been bogus since 1983. Incorrect drivers, user error, viruses, file corruption, improper install are all common "buggy" problems, but have nothing to do with the chip.

Re:We need a better benchmark, and we need it soon

[billcopc]

We all wish. The problem is that numbers are simply a marketing tool. My favorite example is the Pentium 4, starting at 1.5 Ghz, yet in benchmarks it is no faster than a P3-1000. Sure, _we_ all know that, but those senior IT/IS wiseguys don't, they just look at the number and say "This chip is 1.5 times faster than the P3-1000, and it's a Pentium-4, which should be in itself much better than the older Pentium-3. Let's buy this overpriced bitch." That's how things work. That's why every single PC at my place of work is Intel-powered (except mine of course :)

Now I don't hold any grudge against Intel nor AMD, I have an Athlon at work and a Celeron at home and I think they both have a place in the market, but comparing them is like apples to oranges. The numbers mean very little. This Athlon (old slow slot-A core) runs at 700mhz, while my home PC runs at 850mhz. Some benchmarks say they're equivalent, others make my Celeron glisten in the spotlight. You just can't trust any number, not the mhz, not the MIPS crap, not even the independent benchmarks. The only true benchmark is your own feeling. Does this Athlon feel faster than my Celeron at home ? Yes it does. I don't know why, but it just does. That's all I need to know.

Re:We need a better benchmark, and we need it soon

[Dervak]

I think it is futile trying to get a single number to describe the performance of different CPUs, especially as a product of various numbers. Other people have pointed out that what is important is what you use the CPU for. That determines what performance number is important to you.

Think about it like a car. What car has the highest performance? Then ask yourself what kind of performance you mean?

Top speed? Acceleration? Torque? MPG? Range before refueling? Tire grip? Driving characteristics? Ground clearance? Payload capacity? Coolness factor? Envy factor? It all depends on what you want to use your car for (and on your wallet).

It is the same with processors. If you are on a budget and only word-process and browse the web, get the cheapest Celery you can find, if you have a little more dough and want to be able to game some too, get a Duron and a GF2MX card. If you use 3DSMax or pack DivX;-), get a 1 GHz or faster Tbird. If you only play Q3 get a 1.5 GHz P4 and a GF2Ultra, for those important 200 FPS in 640x480... ;-)

/Dervak

Looking for a 1U Duron Server

[Ami Ganguli]

Does anybody know where I get one of these in a 1U server for a reasonable price, shipped within the US? It seems like all the rackmount servers are still Intel-based.

Re:You already have a good benchmark: SPEC

[fatphil]

" - vendor independant " No way Jose! Only the vendors who can cough up the money to participate can claim official SPEC results. Until quite recently, that didn't even include AMD! FatPhil
-- Real Men Don't Use Porn. -- Morality In Media Billboards

Re:I told you so!

[Sloppy]

What does a 100MHz DDR FSB mean? It means that the CPU talks to the chipset at a 200MHz data rate even though the frequency is only 100MHz because it can send two bits of information per clock cycle instead of one.

IMHO, the best way to deal with this is to bring back an archaic-sounding word: baud. It's 100 MegaHertz but 200 MegaBaud. And if someone figures out a magic way to cram more transfers into each clock cycle, then we can talk about the baud measurement going up while, at the same time, avoid lying about the clock frequency.

---

Re:We need a better benchmark, and we need it soon

[e_n_d_o]

The Pentium 4, is faster, and "better" than the Pentium III, but only when software is compiled for it. Go read http://www.tomshardware.com for more info. Tom had a real hard time understanding this at first too, but in the end, he was able to show some more impressive results for the P4 with benchmarks running code compiled for the Pentium 4.

Another advantage of the P4 is that it will scale more easily to higher clocks.

Of course, in the real world, it will be a while before MS can take advantage of this. (Anyone know if gcc can optimize for P4 yet?)
---

You were right the first time.

[yerricde]

Baud is "discrete signals per second." BPS is "bits of information per second." DDR modulation sends two signals on each clock pulse, one on the rising edge and one on the falling edge. Because these are two discrete signals and not multiple bits squeezed into one phase/amplitude signal (as with phone modems), the "baud" rate is the same as the "words per second" rate; however, the parallel bus increases the "bits per second" rate to 64 or 128 times the baud rate and 128 or 256 times the clock frequency.
Like Tetris? Like drugs? Ever try combining them?