The F.E.A.R. Issue

To the readers that pointed out our F.E.A.R. results as being unusually high, we owe you a sincere apology.  When we went back to test Conroe for the second time we re-ran all of our tests to make sure that no mistakes were made.  We caught the Quake 4 issue where Conroe’s SMP performance was understated, and we also discovered a problem in our F.E.A.R. testing. 

While our intention was to test both the AMD and Intel systems at the “Maximum” Computer settings and “High” Graphics settings, only the Conroe system was configured as such.  We inadvertently left the AMD system at a higher resolution (1280 x 960) instead of the default resolution (1024 x 768) when you select the “High” Graphics defaults.  The oversight was entirely our own doing as Intel was not running the benchmarks or configuring them, it simply happened while we were setting up both systems at the same time.  We played with different resolution settings and while deciding that we would go with one, managed to configure the two boxes differently. 

Of course this means that our initial F.E.A.R. tests were incorrect, and below we have the correct results with the settings we intended to run both systems on: 

F.E.A.R. - Average Frame Rate

The performance advantage of Conroe makes a lot more sense now, at 20% instead of 41%.  With performance in Quake 4, UT2004 and HL2 in the 20 - 30% faster range on Conroe, the F.E.A.R. results now make a lot more sense. 

F.E.A.R. - Minimum Frame Rate

F.E.A.R. - Maxmimum Frame Rate

To those who pointed out that even the CrossFire X1900 setup would be more GPU bound at 1280 x 960, you were very correct, our original results were inaccurate.  We do strive for accuracy and reliability in our results here at AnandTech, which is why we went back and retested/confirmed all of our initial findings before bringing you this update.  Aside from the F.E.A.R. and Quake 4 issues that we've since corrected, we found no other performance anomalies in our initial results. 

The Benchmark Issue The DDR2 Issue
Comments Locked


View All Comments

  • BlackLotus2 - Saturday, April 15, 2006 - link

    I actually dont think the Intel results are too impressive :

    - they compare a CPU that is out in 6 months with a current CPU, this does not show a real comparison of products that will compete against each other. who knows what AMD is going to have then

    - the 25 % performance leap is just very relative. if you compare gaming benchmarks with AMD cpus, you can see that for every 0,2 ghz more clockspeed you get 3 to 10 % more gaming performance (3 % on quake, 10 % on farcry), just see here:">
    The test was testing with a dual core 2.8 ghz product, but AMD is already shipping out 3 ghz opterons for a week now, which will already lower intels performance leap as the quake3 benchmarks would be around 3 % better. In 6 months, there will surely be 1 or 2 more new speeds, lowering the gap a lot, giving a total of 10 % more performance than in the _current_ benchmarks.

    - the intel cpu had 2 x 2 mb cache, the amd cpu only 2 x 1 mb cache. 1 mb extra cache gives you something like a 5 to 10 % performance. AMD doesnt need to design a new core to increase the cache size, they still can increase the cache size anytime and still maintain socket compatibility. they can easily catch up here

    - socket AM2 will give AMD an extra 3 % performance

    - intel only ships out less than 10 % in 65nm chips right now, most of their production is still 90 nm. so their advantage to AMD is not that big, AMD will also ship the 65nm chips this year. with 65 nm chips, AMD speaks of a 40 % transistor speed advantage, with such a CPU using less power they can archieve higher clockspeeds. in this test, they compare a 65nm conroe with a 90nm fx62. A 65nm fx62 with 400 or 800 mhz more clockspeed can easily catch up with the conroe. In addition to that, AMD will already ship 65nm chips this year, they ordered semiconductor manufacturer Chartered with the production, so expect the chips already this year.

    - there are already some details around for the new K8L core, which is going to be a real FP monster .. it is said to give a whopping 50 % floating point performance increase to the K8-Core,"> , and it´s only a few month after the release of the conroe. I assume this machine is going to beat the shit out of conroe. I also expect it to come with 2 x 2 mb cache, if intel is producing all their desktop parts with 2 x 2 mb the AMD chips will also, and due to the better scalability because of the on-die memory controller the server market will rather prefer this one (just imagine a dual xeon dual core machine with a memory bus bottleneck). First pictures of the 65nm core for this year also show some interesting details, instead of 3 execution units they show 4 execution units on the die, which might already be a part of a K8L-core. In addition to that, the photo of the die shows something which looks like a new cache area, which might be the new ZRAM L3 cache.

    Personally, I wouldnt wait for a Conroe, I would rather wait for this one (if at all). Also, Conroe is quite half a year away, and the new AM2 boards are out in mid of May. The K8L monster will be socket compatible with the AM2 sockets, so it is no problem to just insert a new K8L-CPU into that socket later, and therefore the chances for upgrades are great, I wouldnt say that for Conroe.

    Intel will try to make most of the money out of it, so I dont expect the chips to sell cheap in the first 3 months. It´s rather price efficient to build and AM2-system right now with the cheaper AMD-cpus, and then next year just put a K8L-cpu into the socket.

    If we sum it up
    - we already have 3 ghz AMD chips, 3-10 % more performance NOW. 1-2 steps coming in the next 6 months, which will lead to 10 to 20 % better performance than the FX62 now
    - 3 % extra performance because of DDR2
    - 10 % extra performance from larger caches if they are introduced yet
    - shrink to 65nm allows higher clockspeeds, gives at least 15-20 % and scales up.
    The same way Intel can increase the Mhz on the Conroe to get more performance, the
    same way AMD can increase the Mhz on their 65nm chips to even out.
    - new K8L core ahead, with 50 % more FP performance (other specs yet unknown, but
    I expect AMD to prepare a beast, they had time enough to design a new core and they
    certainly not only improve the FP performance, I assume the rest of the new core
    to offer a similar performance jump).
  • Rendition - Sunday, March 12, 2006 - link

    With a 4MB L2 cache, won't that increase the price of the Conroe part and lower yields?

    Isn't it likely that we won't see volume production of the Conroe part for desktops in late 2006 due to the large cache and 65nm process?

    If that happens, won't this be like the many paper launches of a high performance product (ATI Crossfire, Nvidia in the past) that doesn't mean much because the vast majority of us can't even buy the CPU?

    I mean we aren't exactly swimming in Core Duo parts (65nm) right now, other than Apple.

    20% advantage from a pre-production processor sample with a HUGE 4MB cache that won't be out for 6 months, isn't that impressive.

    Can Intel actually produce it in volume at a competitive price (with the large cache) in 6 months, is the key question?

    You have put your reputation on the line by proclaiming Intel has taken the performance lead, when you are not even sure if Intel can produce it in volume in the 2H, with enough yields of a super large cache, and along with the fact AMD won't be standing still in the next 6 months.

    I hope for your sake Intel isn't pulling your chain.

    People also have been a little too forgiving on your mistake on the FEAR benchmark. You played right into Intel's hands by posting the bogus 40% performance increase number. It was totally irresponsible and unacceptable to be wrong like that on such an important claim.
  • Conroe - Sunday, March 12, 2006 - link

    no, no, no, and yes.
  • overclock2050 - Sunday, March 12, 2006 - link

    These results are a little strange, too big the advantage of Conroe especially in games.

    I readed also that intel havent permit to control what's was inside the two sistems, this is a little suspect.

    Maybe there's an explanation for these result, "Dynamic overclocking" what everyone considers the results of a 2.66Ghz conroe, may be the results of Conroe running at an hygher clock.

    3.3Ghz or more??? It's only a possible solution, until we will see a real indipendent test
  • JoKeRr - Sunday, March 12, 2006 - link

    wow!!! just wow!! previously we didn't have bios screenshot of FX60-->therefore conroe wasn't running against an FX60--->not fair.

    now it's proven conroe is running against FX60@2.8ghz, so "well we don't have a screenshot of conroe actually running at 2.66ghz" therefore it must be running at 3.3ghz or more!!! Oh man I'm laughing my ass off. I guess some ppl just can't accept the fact that unleashed conroe is faster than AMD's best. Intel definitely needs an excuse just to cheat to win, right?? Give me a break plz.

    as for quantity production of chips on time, AFAIK no one can make chips as fast and many as Intel. Intel was always the first to release and they never had much of a yield problem. (from 180nm->130nm->90nm->now 65nm)
  • theteamaqua - Saturday, March 11, 2006 - link

    its funny to see how Intel beats AMD in 32 bit application, AMD fan boys starting to trash talk about how Conroe's lack of 64-bit support, sure its only memory with bigger register, but it gets the job done

    i doubt the we're gonna move to 64-bit platform within 3 years(i meant every apps, games u see in the store are 64 bit, not just couple of em, but all), there are too many applications written in 32 bit, not to mention games, and the only game i know is Farcry 64 bit, thats it, i think both Intel, AMD , M$ are devising some strategy right now on how to make everyone use 64-bit as we are speaking, but if the software developers don't work with them, they're gonna move very slowly.

    I think every bit generation can maybe survive around 8 to 10 years (thats just my guess based on the old Windows)
  • theteamaqua - Saturday, March 11, 2006 - link

    its funny to see how people react when they knew this was coming, Anand, has been talking about Conroe since like 3rd quarter of last year, and in Tom's Hardware, they did benchmark Pentium M overclocked 2.133 GHz compare to FX 55, it actually beats the FX 55 in some games, the news doesn't suprise me, but again both system were setup by Intel, and until Anand + other review site get their hands on it , I guess this is perhapse the maximum potential the Conroe has, scèrw that , im getting Conroe anyway
  • cornfedone - Friday, March 10, 2006 - link

    Even Supermicro know the truth about Intel.">
  • dev0lution - Friday, March 10, 2006 - link

    I'd be interested to see how a Nvidia-based Intel and AMD platform with the same chips stacked up, as SLI is a more mature platform and in some respects matches up better.

    It's interesting that Intel chose to go the crossfire route, as ATI has been their 3rd party chipset supplier as of late.

    I'll reserve judgement until Conroe is in retail, AM2 is out and I can see benches on multiple board partners, with DDR2 800 and on Nvidia as well ;)

    Love to see thermal and power benches too....
  • theteamaqua - Friday, March 10, 2006 - link

    finally AMD fan boy can shut up about "unknown Processor" issue regarding to the AMD FX 55 BIOS. I think its stupid when they make that comment. However, both the system was setup by Intel, and they didn't let people to do an open box check. There can still be something fishy going on, but I hope not, cause I am gonna get Conroe and if thats the performance i can get it will be awesome

Log in

Don't have an account? Sign up now