HEXUS.reviews :: WORLD EXCLUSIVE! INTEL CONROE BENCHMARKED

[DR]

Wow I can encode a divx 10 seconds quicker, hurrah

I care more about heat/noise and price tbh.

Then the story gets even better....

[htid]

Im definitely going Conroe when they come out, but i did read somewhere (can't remember where) that when they first come out there may be some bugs and compatibility issues, so it would be better to wait a while after they came out until they are more stable. Is there any truth in that, because i dont want to be buying Conroe as soon as it comes out only for it not to work correctly...

[rad]

ok to answer alot of ur questions , intel conroe 2.66 ghz is 529 US , FX 62 is 1219 US , now even if conroe was extremly the same performance its still a spanking cuse the conroe 6700 is HALF THE PRICE! soo any of u amd fanboys tryin to knock conroe , go spend ur extra 700 US dollars on a fx62 , cuse im gannna save that 700 us and spend it on dual gfx cards , i was gnana buy a am2 fx 62 but since intel conroe is coming out very soon iv changed my mind im waiting the extra few weeks and geting a conroe cuse id personaly rather buy a processor dual gfx cards all for the price of a fx62? lol , but seriously all of yous can do anything u want and buy a fx 62 and waste all that money but if ur smart and get over the biasness and stop being amd lover boys , ull save alot of money and have atleast the same performance and some apps ull have alot more but its ur choice just pointing out the plane facts that its not good to be a fan of companys the best thing to do is buy the best products at the time , atm its amd am2 fx 62 but in a few weeks time conroe will b the best choice and its worth the few weeks wait

I agree with what your saying. However I am not too sure the price difference will be that large. Why would a company who already has dominant market share sell something that is so much better than its competitor for so much less. This is intel after all (who have been renowned for selling their products at a premium, even when they didn't have the best cpu to sell).
Anyway I hope your right, because I will be joining you. I just wouldn't hold your breath just yet.
I'll pass judgement when they are in the store.

[Tarinder]

These Conroe numbers sure do look sweet.

I'd love to get one when they're released.

If Willy Deeplung, our intrepid, daring reporter, was able to get some more quality time with Conroe, which benchmarks would you like to see run on it and our comparison FX-62 and Intel Pentium Extreme Edition 965 machines?

[sawyen]

These Conroe numbers sure do look sweet.

I'd love to get one when they're released.

If Willy Deeplung, our intrepid, daring reporter, was able to get some more quality time with Conroe, which benchmarks would you like to see run on it and our comparison FX-62 and Intel Pentium Extreme Edition 965 machines?

LOMAC..

[keta]

yea lusonico i see wat ur saying , but i highly doubt a fx 62 will drop 700 dollars in 4 weeks ? its there premium product soo they charge a the extra money for it , where the 6700 is a mid range conroe , in mid range u got 2.4 2.66 2.93 , 1066 fsb , then in xe u got 3.33 ghz 1333 fsb , the xe will b the top dollar intel product where ud pay the same as a fx62 but this bench mark is comparing a 6700 mid range intel against amds top gun , i really doubt amd is gnana lower there prices from 1200 to 530 in a few weeks just cuse intel have a betta product ? intel neva lowered there prices and amd was betta for 2 years , soo u can keep hoping but the most the fx 62 will drop is 200 dollars to about 1000 maybe 900 us if ur EXTREMLY LUCKY and then still the conroe 2.66 6700 at 529 dollars still looks a betta product for the money , now for u people wondering about heat and power cumpstion , its 65 nm core duo 2 , its very cold and very low power consumption , about the 65 to 80 watt range like amds mid range products. if u still saying amd is betta ur bias cuse this is clearly the mid range intel beating a high end amd , wait till l8 q3 when the high end intel conroe 3.33 ghz 1333 fsb comes out then we will match the fx 62 against that , if this processor with 2.66 ghz 1066 fsb is doing this good , imagine wat that will do ? soo dont keep thinking this product is gnana be expensive cuse its mid range!! not high end so hopefully that has cleared things up for u , but i still have a feeling amd fanboys wont belive it till there on the shelvs in boxes. even that i doubt ull be happy enough to move over to intel lol

[Bob Crabtree]

Conroe E6700 spanks FX-62's botty EVEN MORE FIRMLY than was calculated in the article, in percentage terms.

The percentages of how Conroe performed relative to FX-62 were incorrectly calculated in all cases in which a "lower is better" result is referenced.

For example, in the DivX encode benchmark, Conroe E6700 is 32.20% faster than FX-62 (rather than the 23.78% listed in the article).

The corrected percentages of the Intel Conroe E6700 system's advantage/disadvantage, relative to the HEXUS FX-62AM2 system, are as follows:

-38.02% ScienceMark 2.0 memory bandwidth
- 2.12% ScienceMark 2.0 memory latency
+26.41% HEXUS Pifast
-12.57% HEXUS Cryptography
+19.09% Realstorm Raytracing 2004
+31.20% DivX encode - multithreaded
+32.46% WAV conversion multithreaded
+15.77% CINEBENCH multi-CPU render
+51.32% KribiBench v1.1 - Jetshadow model
+39.47% Far Cry - 1024x768 - speed
+ 3.67% Quake 4 - 1600x1200 - 4x AA 16x AF
+ 0.88% Splinter Cell: CT - 1600x1200 - 4x AA 8x AF

(The corresponding Conroe percentages vs. the Pentium EE 965 are similarly incorrect in the article.)

All of us who've been involved with the testing, the writing and final production of this article are very pleased with the attention that it has attracted and how thoroughly a lot of people, including yourself, have pored over the details - cos, of course, the devil is in the detail.

So, when I was first alerted to your posting, I though, oh bum (botty?), that's MY fault - because I hadn't bothered to check all the figures, believing that was no way that Willy Deeplung would cock them up (stupid, I know - even Willy is human - although the large amount of high-quality work that he can turn out in a short period of time might suggest otherwise).

Fortunately, though, he didn't cock up and, in this instance, it didn't matter that I had not checked his figures - something, though, that I have now just done.

These are the result I get - using the figures shown in the top table on page 3 and bunging them into MS Excel 2000:

For Conroe E6700 vs FX-62 AM2
38.02%
-2.16%
20.89%
12.57%
-19.09%
23.78%
24.50%
-15.77%
-51.32%
-39.47%
-3.67%
-0.88%

For Conroe E6700 vs Pentium EE 965
-11.09%
-35.29%
-38.40%
17.74%
74.58%
-18.30%
-23.49%
19.97%
12.38%
59.99%
7.22%
2.93%

These are - give or take the odd one-hundredth - as per the figures in the second table on page three.

Not sure why some of your figures are so far out and not others - human error I presume - but just so others can, if they wish, check for themselves, the formulas that I used (and presumably Willy, too) are:

(FX62-Conroe)/FX62

(EE965-Conroe)/EE965

In each case, of course, you need to multiply the result by 100 so as to finally express it as a percentage (something that Excel does automagically if the destination cell is formatted appropriately - ie for percentages).

Oh, and for anyone who fancies checking but not retyping all the results, know that if you highlight the top table (the results), you will be able to copy it and paste it directly into Excel (and maybe other spreadsheets, too).

If you paste it into the top-left-most cell of Excel (A1), then this is the forumula that you put into cell G2 (and then propogate it down that column) to get the results for Conroe E6700 vs FX-62 AM2

=SUM(D2-B2)/D2

For Conroe E6700 vs Pentium EE 965, paste into cell H2 (and propogate downwards), this formula:

=SUM(B2-F2)/F2

In each case (in Excel) I converted to percentages by highlighting the relevant cells and clicking the percentage toolbar icon - the alternative is simply to multiply by 100.

Bob C

Update - 19:20

Sod's Law - you stand up in a public place and start singing and realise later that you forgot to put on your trousers!

Translation - there are some errors in this posting of mine and I'm working now to correct them.

Update - 21:46

My posting below - timed at of 09:32 PM - has corrected figures.

These are percentage figures for the second table and relate only to results that were shown in the top table in units of time.

Thanks to mxyztplk for pointing out the problem (and for taking my initial dismissal of it withour rancour)

[dfisher1]

Have you guys (Hexus writers) contacted AMD yet to see what they have to say about their future products.

AM2 was not going to be AMD's saviour. Yes, there is a performance increase compared to pre-AM2 but the increase is actually pretty small. Most people won't be able to tell the difference and even those that do will only find they gain a few seconds or fps.

Come AMD give us some information about your forthcoming products and ideas otherwise some of us will be defecting to the other camp unless we hear something from you.

[balix]

I would REALLY like to see the same comparison with the FX-62 using 1 GB of 667 DDR2. Typically the lower latency rather than high bandwidth will make a huge difference in speed in a lot of applications. Also, the 2 GB rather than 1 GB could alter some of the internal DDR2 timings making it somewhat slower.

[merlin2001]

another way to calc % chgs is [(A/B)-1]*100 which is mathematically identical to [(A-B)/B]*100. So anyone uses the 1st, bob's math is correct in that sense...
***on which is which
if you are looking to see the % differance between two objects A and B where you want to say that A is x% differant to B, then the maths is above.
If you say The percentages of how" Conroe performed relative to FX-62 ". then the maths is (C - FX / FX) of course, if the article reads "how FX performed compared to/relative to Conroe", then you should do the maths: (FX - C/C). I haven't dble-checked which way the article calls it, but hope this is clear.

btw. - why use =SUM in your formula?

[t4d]

To say that "Intel's Conroe spanks AMD FX-62's botty" is something of an over statement. I have been reading the Hexus reviews for sometime and had in the past been impressed with their non-biased results, but this one has to be a joke.

The figures certainly seem impressive at first glance, but then there are so many anomolies in the comparison between the two systems used that I can't believe that this review could ever be considered valid.

1) Using a Foxconn motherboard for the FX-62 (AM2) with 2GB Ram alongside an Asus mobo for the FX-60 (939) and all the other setups running with only 1GB ram. Not to mention running the Intel Conroe on the manufacturers own Intel motherboard

2) The Benchmarks - Using Hexus' own inhouse benchmarks seems good on the suface, but are they optimized for Intel or AMD, both or neither?

Using the Realstorm 2004 benchmark which is 2 years old for new technology?

3) Running Farcry at only 1024x768 whilst running Quake and Splinter cell at more demanding settings.

4) Almost forgot, lol, What's the deal with running the tests for 64-bit processors on a 32-bit operating system?

So come on you guys, do us all a favour and do a fair review with a consistent setup and benchmarks.

Intel may have done a good job and may indeed have released a processor that is faster than AMD, hats off to them, but let's not forget that AMD is still currently the underdog here with a smaller market share and that it was AMD that brought us the dual core processors becuase Intel did not believe that consumers needed them.

[Dianoda]

I agree with what your saying. However I am not too sure the price difference will be that large. Why would a company who already has dominant market share sell something that is so much better than its competitor for so much less. This is intel after all (who have been renowned for selling their products at a premium, even when they didn't have the best cpu to sell).
Anyway I hope your right, because I will be joining you. I just wouldn't hold your breath just yet.
I'll pass judgement when they are in the store.

The reason why Intel would charge a much lower price for a better performing chip is because it can. Intel wants market share back from AMD. And it's pretty much the perfect storm for Intel, they have a product that is better than the competition (or so they believe, and so do I), and can make money on selling those chips at a much lower price than the competitor's equivalent product, and AMD simply can't compete with a 90nm process cpu, and won't be able to until the end of the year. If AM2 was the debute of 65nm process AMD chips, then it's chips could compete because higher clock speeds could be set with stock cooling. AMD is supposedly going to break out 65nm AM2 chips at the end of the year, which should OC like crazy, and push AMD into 3.4GHz+ territory easily, with plently of headroom, too. The 65nm Prescotts did the same thing for Intel a little while back, stuff like a 2.8 Pentium D running @4.25GHz on air, imagine what AMD could do with that. K8L is still pretty far off, AMD will compete in the short term with 65nm K8s. Which probably couldn't come soon enough for AMD.

[mxyztplk]

Bob,

The problem is that the experimental results are expressed in different terms: some are in units of output PER (i.e., divided by) unit of time ("higher is better'), and others are in units of time ("lower is better"). The results expressed in units of time must be inverted (to express them in units of output per unit of time) before computing the relative performance. Otherwise, one is computing based on inconsistent measures in one case vs the other.

If the time figures are first inverted, a single formula can be applied to get the percentage advantage/disadvantage of Conroe vs. FX62: 100 * (Conroe / FX62 - 1)

Or, if you do not wish to invert each time figure, you can use two different formulas:
1. For results using units of output PER unit of time ("higher is better"), the correct formula should be: 100 * (Conroe / FX62 - 1)
2. For results using units of time ("lower is better"), the correct formula should be: 100 * (FX62 / Conroe - 1)

[In formula #2, the term "FX62/Conroe" is simply equivalent to "(1/Conroe)/(1/FX62)" ]

To give a simple (though extreme) example, if Conroe ran a benchmark in 10 seconds and FX62 ran it in 20 seconds, Conroe would be 100% faster than FX62: 100 * (FX62/Conroe - 1) = 100 * (20/10 - 1).

Alternatively, if the result were expressed in "ExampleMarks" ("higher is better"), Conroe achieves 0.1 ExampleMarks, vs FX62's .05 ExampleMarks. Applying formula #1, we get: 100 * (.1/.05 - 1) = 100%. This is the same result as formula #2, which is what it should be.

[DR]

This is not a final review; this is a preview trying to let people know the information.

We had constraints (read the thread) I am sure you will find our final review more final.

I am also shocked how you are debating Bobs calculations (not even) the ones which the editorial member used. He just did it to prove percentages are correct.

We are comparing a mid range part with a new high end AMD part - it does spank it, plain and simple.

DR

[DR]

I would REALLY like to see the same comparison with the FX-62 using 1 GB of 667 DDR2. Typically the lower latency rather than high bandwidth will make a huge difference in speed in a lot of applications. Also, the 2 GB rather than 1 GB could alter some of the internal DDR2 timings making it somewhat slower.

I am sure we can do this with other chipset reviews for our final piece...

[mxyztplk]

David,

The percentage relative calculations for those results expressed in units of time ("lower is better") are wrong in the original article, for the reason I gave. (I addressed my explanation to Bob, inasmuch as the author of the original article did not comment on my post, whereas Bob did.)