CAN YOU BELIEVE IT? INTEL ACTUALLY GAVE US A DUPE BENCHMARK!

[TabrisDarkPeace]

How not to compare a single GeForce 6800 GS to Crossfire Radeon X1900 XTX (see above):

The facts to look for before posting and looking the jackass follow - They are suggestions for some people to consider.

Did they even run the same resolution ?

Was FSAA was used ?

Which filtering type was used (eg: Trilinear vs Anisotropic 8x) ?

Does Unreal actually run faster when in Crossfire mode ? (most likely it does, and the settings and resolution were set higher than TomsHardware CPU tests, which... being CPU tests are run in basic settings at a basic resolution).

Please check your (non-)facts before posting.

As for the foolish BIOS version comments above, I suggest readers just ignore them, pretty much every BIOS on every board on the market is using an AWARD v6.00 as the source BIOS, which is then modified by the vendor (eg: DFI, Abit, MSI, etc) , Copyrighted until 2003, clearly not a version indicator.

However, the actual board / vendor version might be 1003, or 1008.003 beta (eg: Asus), or other (eg: Bios ZM is later than YB as Abit used to use).

Applying the latest BIOS to that DFI board gains you under 5 fps in most of those tests, the tests are not fakes.

Do the creators of the Award BIOS as a company still exist ?, Did they recently change name to or from anything ?, When you can answer that maybe you'll have a basic understanding in BIOS versions 101.

Much like a Intel Celeron 466 outperformed the AMD K6-2 550 in Quake 2 (one of the 3DNow! optimized games at the time), the Conroe at 2.66 GHz outperforms the Athlon 64 at 2.80 GHz. They've done it before and they'll do it again, it is just a game of chess or cards really.

This is normal as Conroe has 4/3 higher IPC, as it has a 4-issue core, vs the Athlon 64' 3-issue, vs the Pentium 4' 2-issue (although core design made it more like 2.5 at times) cores.

Do people here even know how to compare say a hypothetical 16-issue processor with a 16 stage pipeline clocked at 1 GHz to an existing processor ? (eg: How would it perform compared to a hypothetical Athlon 64 core clocked at 4 GHz ?) (all other core elements being equal, or proportional where required).

When AMD release a new K9/K10 core with 5/3 or 6/3 (double) the IPC of the current generation Athlon 64 you'll understand CPU architecture 101 a little better (I hope).

When I get my Conroe I'll complain here my MINIMUM frame rate is too high and ppl will suggest I get an Athlon 64 to slow it down (by comparison) as a joke most likely. 😛

PS: No fanboy here, look at my sig, currently quad-core Opteron 270, doesn't exist ?.... well it does with 2 CPU sockets each with 2 cores ? (Is is sinking in yet ?). Sad because the investment is not looking as good as anticipated (got it around the time the Pentium-M / Centrino came out in Australia, not much to compare it too, including the leaked roadmaps on Wikipedia).

 

[noblekitty]

It is funny for the people who love AMD to call all others as fanboys of Intel. But if they just take a few minutes to read over all their past posts, they will see who are the real fanatics here.

I just join this forum recently ( though I've been a readers of Tomshardware.com for more than 5 years now), and the bad impression I gotten has to go to the "AMD fanboys". They always act like a bunch of kiddos. Call me an "Intel fanboys"? feel free to do so. But I am currently running a rig with Opteron 148 ( oced to 3.0 ghz) and loving it for gaming. I also am running a Pentium 640 (oced to 4.0 ghz) rig; and I love it for the ease of choosing softwares in video encoding/editing ( i am doing a lot of divx encoding).

I will use both Intel or AMD cpus as long as they are good and reasonably priced. Conroe is Intel's hope to regain the ground they been losing to AMD for a long time now. I really doubt Intel will use tricks or lies to gain a 20% to 40% performance. If it was only a 5-10% gaining in performance, I would have doubt. Plus, the folks who support Conroe are those who embrace technology greatness, not neccessary an Intel fanboy. They formed their opinions on what Intel had shown them. And if 6 months from now, It turns out that Intel have been lying to us; I am sure it will do no good from them from where they standing now.

My only concern now is that if AMD has nothing to compete with Conroe, Intel will enjoy pricing their cpus anyway they like it.

Did anyone else notice that since the day AMD was doing good with their AMD 64, their CPU pricing have been relatively high against Intels. AMD is no longer a budget CPU brand like they used to be with Athlon, K8, K7... 😀

 

[the_guru]

Did they run the same time demo?

No they probably didn't since THG has their own time demo. So I guess I was a bit too quick... Anyway I think people should wait before coming to conclusions.

 

[Action_Man]

Exactly wait till we have independent benchmarks and then we can say whether it was rigged or not. Everyone get that?

 

[TabrisDarkPeace]

i really wonder... if memory latency isn't that important, then why corsair actually have a lot of demands for their ultra low latency DDRs?

It is because they mass advertise, and the ignorant masses believe it still makes over 5% difference like it used to back in the days before DDR and integrated memory controllers. (Thus they find documentation on it and the 'outdated facts' indicate it does).

Tell me do you drink Coke Cola or no-brand Cola ?
- Coke advertises very well and charges 4x what their product may really be worth.

Low RAM timings still help, but nowhere near what they used to.... of course high RAM timings are still a bad thing, but there is a very cost & performance effective middle ground often ignored (eg: G.Skill).

Tip: Overclockers are better off with high clock speed / typical timing RAM than normal clock speed / ultra low timing RAM.

 

[TabrisDarkPeace]

l
call DFI, and ask them if they've delivered any mobo with 2003 bios this year?

Just checked 3 different DFI boards all from this year out back, and guess what ?

They all use Award BIOS v6.00 as their 'source' BIOS.

Sure I can upgrade the BIOS revisions on the boards, but performance is highly unlikely to magically jump more than 5% in the tests on a similar system. Quite often BIOS updates actually decrease performance by fractions of a percent... but yes I agree this one will raise performance a little.

Viperabyss, unless you actually have facts do us all a favour and avoid posting in future.

PS: Can I bill you AU$100 for my time and spreading misinformation to my potential future Conroe clients ? (this thread will still exist in 6 months, unless deleted by a moderator).

Intel listened to what the market wanted:
- Higher IPC at similar clock speeds
- Lower power consumption (esp compared to PreScott)
- Greater than 3-issue processor cores
- Faster processor to system / memory interface
- Multiple cores with a shared L2 cache that do not communicate via the external FSB.

Now they've delivered this and your 'little world' is falling apart.

Welcome to the world of IT, over the next 15 years you'll see stuff like this quite often.

I've recently (well not so recently) purchased a rather expensive Opteron 270, (4 x 3-issue cores) and the Conroe will be supplying 2 x 4-issue cores (per socket), which treads on my systems performance and the CPUs will certainly cost less than AU$3,200 all up (what I spent on CPUs alone not that long ago).

Think of it like this:
- Pentium M and Athlon 64 are, clock for clock, very close in performance
- Intel Core Duo is already providing very similar IPC to the Turion 64 X2 in performance, and has been out longer (laptops are the largest growing market segment).
- It is easy to conclude that Intels' (still to be named) 'Conroe core' would have even higher IPC again than the Pentium M, while being far more desktop oriented and clocked at a higher speed (partially because of 65nm manufacturing process transition).

For gaming AMDs upcoming 'quadcore on a chip' for AM2 will perform slower than Intel just having 'dualcore on a chip' (remember: each new Intel core is going to be at least 4-issue from now on), as when creating games (excluding dedicated game servers) it is very difficult and costly to have their performance scale on quad processor machines. Game creation deals with 'what will be the norm at the time of release' as 80% of their buyers will have 'typical' PCs of that era.

ie: You'd need to clock an Athlon 64 (Socket AM2 65nm varient) at almost exactly 3.72 GHz just to break even with a 2.66 GHz Conroe.... the 65nm transition will not let AMD get to 3.72 GHz with normal cooling within an acceptable power envelope at the same price / performance ratio Intel are likely to offer on this (they hate losing market share, when they do they fight back with avengence).

The only way AMD can counter is:
- Make every game 100% quadcore optimized for their architecture (and god I hope they do, as running Opteron 270, but we all know they wont)
- Release a 5 - 8 issue processor architecture at similar clock speeds to Conroe with only 2 cores per socket but shrunk to 65nm.... ala: AMD K10 or higher. A few people in the know have been aware for some time that AMD K9 is going to be skipped and was just an internal testing design some time ago.

Of the two above 'counter offensives' the lower is actually more likely to be successful,... assuming they'd kept some R&D up their sleaves.

 

[Guest]

Fuck all the idiot fanboys.

Just wait; buy; test yourself. Until the STFU!

 

[ToxicCow]

AMD has not been trumped. Intel is basically saying "A product we are going to release 6+months from now will be better than products that are available now" DUHHHHH. When conroe comes out and if it outperforms AMD's, then I would imagine AMD would make a better product that can beat conroe sometime after.

And anyway, why on earth would anyone care who's product is best just so long as you get a good deal?

Really, if conroe performs well, I will be very happy since the consumer gets a good chip and it will inspire AMD to release even better chips sooner.

Conroe = Good
Consumers (should) = Happy

Conroe = Bad
Consumers = Unhappy

 

[ltcommander_data]

I originally posted the following in another thread, but I never recieved a satisfactory response.

Now, I know you didn't write the article, but I'm wondering how Mr. Sood knew that the DFI motherboard was using the D49C-32 BIOS version. His entire argument centres upon Intel using that BIOS version which didn't include many fixes that would have affected performance.

He saids:

Quote:
You’ll notice that the image I am referring to on Anandtech's website (the bios image) states that the AMD processor is “unknown” which makes me believe that the bios they are running is outdated. So, I did a bit of digging and low and behold, the DFI bios version “D49C-32” they are running is from 10/11/05.

However, I click the link he gives and I don't see D49C-32 mentioned anywhere. The BIOS image that Anandtech provides only states the AwardBIOS v6.00PC is included which is not the actual BIOS version. It's simply the standard BIOS core used by most manufacturers who then costumize for the specific motherboard. Maybe I'm not reading the image properly?

Interestingly, when I type D49C-32 into Google nothing comes up.

http://www.google.ca/search?hl=en&q=D49C-32&meta=

According to the DFI website there are only 2 BIOS versions available the RDXDC23 and the rdx2a11.

http://us.dfi.com.tw/Support/Download/bios_download_us.jsp?PRODUCT_ID=...9&C

Now Sood specifically claims that the BIOS Intel was running was the D49C-32 so I think he's probably referring to the RDXDC23. The funny thing is the RDXDC23 was released on 2005/12/23 and includes all the fixes to the problems that he claims the AMD test system had. If we go by date, the only BIOS released on 10/11/05 was the rdx2a11 whose name looks nothing like the D49C-32 that he's referring to. I haven't actually tried installing the BIOSes on the website since I don't have the correct board, but I'm assuming their file names are representative of their contents.

Maybe I'm seeing things or overanalysing the situation, but if you could point me to the image on the Anandtech article that Sood said showed that the motherboard was running the D49C-32 BIOS version that would be appreciated. If you can also clear up the discrepancy of just which DFI BIOS version that the Sood is saying Intel was tricking us with that would also be nice.

 

[edwuave]

intel fanboys shouldn't be too proud or snobish, but sit and wait the real benchmark in the next 6 months. and AMD fanboys should celebrate since conroe means a birth of a more powerful processor(than conroe) will be unveiled soon after conroe's release.

fair and square, there shouldn't be any arguement here. it just merely a benchmark of an unknown AMD processor agaisnt intel's superprocessor. the real benchmark will unveiled everything. even if the real benchmark indicate that AMD lost to conroe's performance by as much as 20%, it means nothing-does it mean we should buy conroe right away?it will burn a hole in your pocket.


http://voodoopc.blogspot.com/2006/03/if-only-they-had-time-machine.html


in the mean time, we consumers are the winner actually. cheers.....

 

[TabrisDarkPeace]

Sood is the CEO of VoodooPC

Of their 8 PC lines, 7 are AMD processor based. (87.5%)

One is a using a mobile Intel processor:
http://voodoopc.com/showroom.aspx?lineID=17

CEOs are meant to protect their companies interests, and Soods interest is making profit by selling AMD processors.

VoodooPC have invested considerable effort to support AMD, having them 'defect' to Intel would have a negative impact on their public image and lower their 'gamer respect'. [Fist on chest, zooms out to show AMD logo t-shirt - if you get my gist].

Anyone who thinks that Sood has a non-biased view, when he is the CEO of a company selling 87.5% AMD machines (more like 97.3% by volume of VoodooPC sales) is a fool.

Do AMD fans actually purchase machines from VoodooPC, or build themselves using their technical know how ?

Sood appears concerned at the least that VoodooPC (his company) will not be able to source Conroe (or any Intel) processors 'in quantity' and need to slash sale prices (and thus profits) to compete with other 90nm and 65nm AMD machines on the market.

 

[gudodayn]

People that say Intel has no reason to lie, what are you talking about?
Is this why Intel made up all the BS with Skype?

They have a lot hanging on the line as AMD is improving their machines as well and the server end with the Opteron performing better than Intel.

To me, the only real advantage Intel have is in the notebook sector. If I was to buy a notebook today, I'd get one that runs Intel Core Duo, no doubt.

But you know what, its Intel! It's just what they do! They like to say things in their favour!
Not to say that Conroe is going to suck but when the final product comes out in 6 months time, it better be up to scratch to compete against whatever AMD has to offer in 6 months time!!!!!

For now, can Intel grow up and show customers something SOLID!!!!

 

[Comas]

Intel should stop fucking around with these results. Just shows how much they will do just to take the performance crown back from AMD. Hope it never happens, just a pity to buy processors from a company like INTEL.

 

[Action_Man]

People that say Intel has no reason to lie, what are you talking about?
Is this why Intel made up all the BS with Skype?

:roll: I don't have the words.

They have a lot hanging on the line as AMD is improving their machines as well and the server end with the Opteron performing better than Intel.

Woodcrest topped Opteron.

 

[Rabidpeanut]

This is what happens when you become a fanatic, i like amd and i am dissappointed that they (possibly) might lose this one. But no one knows yet so you should all put down your respective brand name genetalia and wait until you have something real to fight about.

 

[TabrisDarkPeace]

Updated review covers all the issues above in great detail:
http://www.anandtech.com/tradeshows/showdoc.aspx?i=2716

- As explained above the Athlon 64' RAM timings made a very minor difference
- It was also found the Conroe was using 5-5-5 timings, originally thought to be 4-4-4 (typical of DDR2-667), they also have been corrected
- The Quake 4 benchmarks in particular have been updated, but so have all the other tests.
- Action_Man (above posts) has quite correctly pointed most of this out to you in black and white terms.
- Hopefully all the ****wits out there will now notice that Cool'n'Quiet actually saves power when idle, but does not hurt performance very much at all (assuming rest of the system is setup correctly of course 😛)

CeBIT Germany started on the 9th of March 2006, and ends on the 15th, so it will be interesting to see how AMD responds during their 'hometown' (well home country in Hannover) consumer electronics exhibition right after the Intel Developer Forum (IDF) ends.

Because I am so nice:
http://en.wikipedia.org/wiki/Hannover

By the 1st or 2nd week in April (as usual, every year) we'll know how the rest of the year is going to pan out, on a month by month basis with a very high accuracy.

Tip: Mid April is also a good time to buy video cards, and has been pretty much every year since the AGP slot was introduced. Although you may miss the 1st rev DX 10 cards (unless you're rich) later in the year... but most smart techs would want to wait for 2nd rev DX10 cards in April 2007 anyway 😛

 

[Ycon]

Check AnandTech, they re-did the benchmarks with a new BIOS version.
Now it does recognize the FX and as you expected, the performance doubled.

 

[TabrisDarkPeace]

Now it does recognize the FX and as you expected, the performance doubled.

Don't tease them, some retard will believe you w/o checking the link to the updated Anand article (above). 😛

AMD has not been trumped. Intel is basically saying "A product we are going to release 6+months from now will be better than products that are available now" DUHHHHH.

The Athlon 64 FX-62 is available now is it ?, Please provide me a link to where I can purchase 100's of them offically to resell immediately. 😛

Back to reality:

It is a shame in some ways as the German economy needs AMD to succeed, although CeBIT will get them many visitors (each bringing money to the country)..... I still hope AMD can hold out to 'keep the balance' and keep price/performance ratios good.

When I first got my Tyan K8WE (S2895 varient) the board had no idea about the Opteron 'Italy' core, and saw 4 processors when it only expected to see 2 at boot, couldn't id them either. 😛

Performance didn't rise much with the 1.02 BIOS update either (and with BIOS updates it rarely does)

eg: Usually for Asus BIOS updates around the 1005 to 1008 version you gain stability, as some overzealous settings are corrected, and performance drops a shade to help the 'less savy' overclockers get it working right within 10 attempts. If they did this from the word go their sales would be lower, as the board looks 'less bleeding edge'. (Ironic no ?, not to bash Asus as many others do the same thing).

 

[Ycon]

It is a shame in some ways as the German economy needs AMD to succeed, although CeBIT will get them many visitors (each bringing money to the country)

I know its off-topic, but nothing can save the German economy.
And AMD will be able to price their products however they want because they are totally unaffected by the German economy because they dont pay ANY taxes AT ALL and also get major discounts, for example at the insurances (for the workers).

 

[TechnologyCoordinator]

For the love of goodness, can someone benchmark an FX-60 already and see if they get something close to the numbers they got?

Although we don't have a Conroe to benchmark we could at least see if there FX-60 benchmarks are correct.

 

[TabrisDarkPeace]

You are a wise one,

Germany lacks the resources of the USA / China, they can get them, but at an inflated cost of course. 😛

All the 'assistance' they provide to AMD might actually be hurting their country..... I would still hate to see AMD go after all these years, people just like the underdog, and even when running my Celeron 466 @ 583 (Basically a P-II, MMX core w/ full speed L2 cache), I respected people with AMD K6-2/III/2+ machines at 450 - 550 MHz at LANs, etc... for keeping the balance there. (If they didn't I might not be running my current rig - see sig 😛 ).

For the love of goodness, can someone benchmark an FX-60 already and see if they get something close to the numbers they got?

Although we don't have a Conroe to benchmark we could at least see if there FX-60 benchmarks are correct.

Readers, keep your eyes on:

- http://www.amdcompare.com/us-en/desktop/
- http://www.amdcompare.com/us-en/opteron/
- http://multicore.amd.com/
- Note: The FX-62 is not actually out yet either, we are comparing 'future apples' with 'future apples' here... the chipset and memory controller may change things, but not by more than 10%, and more likely by under 5% (for those of you speed reading the rest).

We, and many other hardware sites, already have (see comment above about the BIOS only boosting performance just shy of 5% far above, from me 😉, Anandtech had the exact same results with the new BIOS as here, and other sites did), but we don't have the exact same Anandtech timedemo for Quake 4,... that and the the Athlon 64 FX-60 is only going to give around 92.85% of the FX-62. (The FX series is not multiplier locked, so they set it up exactly how an FX-62 would be setup... although they'll have Socket AM2 FX-62 in 6 months but it'll still perform roughtly, most likely within +-3%, as the setup used).

Remember we are looking at the apx (but very damn close) FX-62 performance (2.80 GHz), not the FX-60 (2.6 GHz), compared to the Conroe at 2.66 GHz. Anand still forget to attract attention to the fact that higher model Conroe CPUs are very likely to have a 1333 FSB (333 QDR) instead of the 1066 FSB (266 QDR) by October when released on newer Intel chipset platforms (vs i975).

Athlough it is true that both platforms are expected to gain memory subsystem performance:
- The FX-62 from Socket AM2 and DDR2-667 (vs S-939 and DDR1-400)
- The higher model Conroes from 1333 FSB (vs 1066 FSB 'choking' the I/O under some uncommon circumstances), and newer chipsets. (AMD memory controller is integrated so newer chipsets from Via, nVidia / ULI, SiS and ATI don't help it very much performance wise).

; Thus it is almost certain that they may both just gain another 3-5% in 6-8 months time, and be in pretty much exactly the same situation as they are today.

This happens every March - April each year, and people refuse to believe the results until October - November the same year. 😛

Unless AMD have a large stockpile of 65nm processors ready (in Oct - Dec '06) they are going to have a tough time keep market share where is it, let alone increasing it, when Vista launches around October.

Obviously Intel timed this to inflict maximum damage, and who can really blame them after having to release the Pentium 4 PreScott (P4 500 - 600 series) and Pentium D (PD 800 series) the way they did ?

 

[RichPLS]

Some have tried to compare the results from these benchmarks to other results, using similar applications but different workloads. For example, our iTunes 6.0.1.3 test uses an input file that’s around 1/2 the size of the one Intel supplied us for these tests. The results in the game and encoding benchmarks are simply not comparable to anything outside of the two systems we have here. These results are not meant to be definitive indicators of performance, but rather a preview of what is to come.

http://www.anandtech.com/tradeshows/showdoc.aspx?i=2716&p=1