1. Welcome to TechPowerUp Forums, Guest! Please check out our forum guidelines for info related to our community.

Don't wait for new Nehalem processors. Read why.

Discussion in 'System Builder's Advice' started by toloratedmeat, Oct 11, 2008.

Thread Status:
Not open for further replies.
  1. toloratedmeat New Member

    Joined:
    Oct 11, 2008
    Messages:
    34 (0.02/day)
    Thanks Received:
    6
    **UPDATED** Don't wait for new Nehalem processors. Read why. Screenshots Posted!!

    The new Nehalem processors' main feature is that is does not have an FSB rather a memory controller onboard. "Bloomfield" the first Nehalem release will be running on an Intel X58 chipset. This requires Triple Channel DDR3 RAM. So this is supposed to "trick" system builders. So it's kinda confusing but i will try to explain it as best as I can.

    So our theoretical mobo has 6 slots.

    [​IMG]

    Here are our slots, the ones with dashes under their name means they're in the same channel.

    So if you had 6x1GB ram you would put one DIMM in each slot.
    If you had 3x2Gb you would put them in 1,3,5 correct?
    In dual channel if you had 2x4gb you would put them in 1,3
    But in triple channel, you would put them in 1,2 because if the third slot isnt occupied, apparently it dosent run as fast as it should.

    Also even though the Bus speed is 4.8GHz/s it isn't as good a gaming processor as a workstation processor. Why?

    The "Bloomfield" series have 4 physical cores and 2 threads each. Equaling 8 threads.
    Most games are single threaded meaning that in the Penryn series more GHz+FSB+Cache meant speed. This here is not the case. Since there are so many threads, this will slow the game down a lot.

    Here are my two benchmarking systems:

    Core i7 Machine:
    Intel Core i7 940 @ 3.2GHz on air
    12GB DDR3 1066
    ATI HD 4870 X2
    MSI Eclipse
    32GB SSD

    Core 2 Quad Machine:
    Core 2 Quad Q9650 @ 4GHz On liquid
    8GB DDR3 1600 Crucial
    ATI HD 4870 X2
    ASUS P5Q3 Deluxe
    750GB HDD SATA II

    In Crysis:
    At 1024x768:
    Core i7 got: 36fps MIN 74.6fps MAX
    Core 2 Quad got: 43fps MIN 88.4fps MAX

    At 1280x1024:
    i7 got: 24fps MIN 53.9fps MAX
    C2Q got: 12fps MIN 65.5fps MAX

    At 1920x1200:
    i7 got: 9.3fps MIN 33.6fps MAX
    C2Q got: 18fps MIN 51.6fps MAX

    As you can see, the fps gap at smaller resolutions are smaller. In the first one, it was a matter of a few frames. But as the going got tougher, the i7 skipped out. I also realised in task manager that while I was running the benchmark one thread was on 100% and the others were on 3-7% idling doing nothing.

    It will be a matter of many years before games are multithreaded. Until then stick with your C2Q.

    The graphics card in both machines were the same. Not overclocked. Stock.

    Thanks to iSkytech for letting me borrow the processor/mobo.
    The Core 2 Quad Gaming Rig is mine so if you have any questions on overclocking just post them here.

    Thanks!!

    Attached Files:

    Last edited: Oct 20, 2008
  2. Darknova

    Darknova

    Joined:
    Nov 8, 2006
    Messages:
    5,037 (1.80/day)
    Thanks Received:
    535
    Location:
    Manchester, United Kingdom
    Run the benches with the CPUs at the same speed with the same cooling method, and use the SSD on both machines. Otherwise this means absolutely nothing to me.
  3. Mussels

    Mussels Moderprator Staff Member

    Joined:
    Oct 6, 2004
    Messages:
    41,943 (11.76/day)
    Thanks Received:
    9,364
    "The "Bloomfield" series have 4 physical cores and 2 threads each. Equaling 8 threads.
    Most games are single threaded meaning that in the Penryn series more GHz+FSB+Cache meant speed. This here is not the case. Since there are so many threads, this will slow the game down a lot."

    yeah yeah, lets all go back to single core CPU's. thats a pretty poor argument there.
  4. oli_ramsay

    oli_ramsay

    Joined:
    Sep 22, 2007
    Messages:
    3,476 (1.40/day)
    Thanks Received:
    569
    Location:
    UK
    Do you work for AMD?
  5. xylomn

    xylomn New Member

    Joined:
    May 13, 2006
    Messages:
    661 (0.22/day)
    Thanks Received:
    18
    Location:
    Swindon, England, United Kingdom
    You can't run benchmarks on cpu's for comparison with completely different hardware on each side, simply unscientific.

    - Same amount of memory @ same speed
    - Same hard drive
    - as your comparing architectures both cpu's @ same speed

    how can you expect your results to be respected when you have missed the above :shadedshu
  6. btarunr

    btarunr Editor & Senior Moderator Staff Member

    Joined:
    Oct 9, 2007
    Messages:
    28,201 (11.41/day)
    Thanks Received:
    13,572
    Location:
    Bangalore, India
    I still feel if you have a Core 2 Extreme, even a QX6700, you can hold on for a while post Bloomfield. After Bloomfield arrives, you'd be forced to sell your ~$800+ CPU at dirt cheap prices...let's say a QX6700 goes for $280, since if you're unreasonable, people would rather buy a new i7 920. So after selling your 1337 chip for a quarter of its original price, you'd be spending an extra ~$600 for a 96x XE, and for performance increments that aren't 'revolutionary', just a little significant.
    CDdude55 says thanks.
  7. Jansku07 New Member

    Joined:
    May 24, 2008
    Messages:
    171 (0.08/day)
    Thanks Received:
    24
    Location:
    Finland
    Second post - no pictures or CPU-Z validation whatsoever. I dont find this results very trustworthy..
  8. toloratedmeat New Member

    Joined:
    Oct 11, 2008
    Messages:
    34 (0.02/day)
    Thanks Received:
    6
    Ok... I'll post the new benchmarks with the processors at the same speed next Saturday because I have school this week. I'll include the CPU-Z stuff too. I need to run back down to iSkytech and I cant do that on weekdays.

    I promise I'll get back to you guys soon.

    Since the Q9650 is stock at 3.00GHz, what should I set the FSB to to make it 2.93GHz. I just dont wanna do the math.
  9. Darknova

    Darknova

    Joined:
    Nov 8, 2006
    Messages:
    5,037 (1.80/day)
    Thanks Received:
    535
    Location:
    Manchester, United Kingdom
    Run both at 3.2Ghz. Most games are single-threaded, so that will show which has the better architecture and how much of a performance increase we can expect.
  10. r9

    r9

    Joined:
    Jul 28, 2008
    Messages:
    2,144 (0.98/day)
    Thanks Received:
    284
    +1
    From this we concluded that higher frequency means more speed daah.
  11. SimFreak47

    SimFreak47 New Member

    Joined:
    Sep 14, 2008
    Messages:
    557 (0.26/day)
    Thanks Received:
    37
    Location:
    In a city, thats in a state which is in the US, an
    As others have stated, you NEED to have the same RAM, same HDD, and same CPU frequency. You also need CPU-Z screenshots, and GPU-Z screenshots. Till then, I will have to disregard this info, just like an above member.
    Edit* What OS are you running?
  12. toloratedmeat New Member

    Joined:
    Oct 11, 2008
    Messages:
    34 (0.02/day)
    Thanks Received:
    6
    I cant do same ram because MSI Eclipse is Triple Channel and unless I have 3 DIMMs of 2.6666667 ram it wont work at full speed. So what should I do?
  13. Darknova

    Darknova

    Joined:
    Nov 8, 2006
    Messages:
    5,037 (1.80/day)
    Thanks Received:
    535
    Location:
    Manchester, United Kingdom
    I would run 2 different benches. One with the i7 in Dual Channel, and one with the i7 in triple channel. With the dual channel run the same amount as the C2Q (Say 4Gb) and then with the Triple run the closest you can (say 6Gb).
  14. Bluefox1115

    Joined:
    Nov 12, 2007
    Messages:
    751 (0.31/day)
    Thanks Received:
    48
    Location:
    RI, USA
    yeah. delete this thread, and start a new one when you can compare with same/similar hardware, and actually get benches proven. :)
  15. CrAsHnBuRnXp

    CrAsHnBuRnXp

    Joined:
    Oct 19, 2007
    Messages:
    5,451 (2.22/day)
    Thanks Received:
    635
    Fail
  16. Darknova

    Darknova

    Joined:
    Nov 8, 2006
    Messages:
    5,037 (1.80/day)
    Thanks Received:
    535
    Location:
    Manchester, United Kingdom
    Not helpful. This guy has access to the hardware I can guarantee we are all waiting to see, so instead of making inane comments like that maybe you should be helping him come up with a benchmark that will actually show off i7 on a level footing.

    I for one would love to see what i7 can really do, but if he goes somewhere else because he feels too "unwelcomed" then you only have yourself to blame.
  17. tkpenalty New Member

    Joined:
    Sep 26, 2006
    Messages:
    6,958 (2.44/day)
    Thanks Received:
    345
    Location:
    Australia, Sydney
    So mate, is it really that difficult to validate CPUZ? I REALLY doubt a school student would have access to this sort of NDA-covered hardware; the Core i7s, as well as information. I'd expect a seasoned engineer or journalist, not you, not to offend.

    In the time you made that post toloratedmeat you could have at least taken us a screenshot. Sure, you COULD Be one of those underage IT prodigies, who works at top-end jobs without qualifications (i know one), but don't give us an excuse like "i had to return the system"
    Last edited: Oct 12, 2008
  18. Darknova

    Darknova

    Joined:
    Nov 8, 2006
    Messages:
    5,037 (1.80/day)
    Thanks Received:
    535
    Location:
    Manchester, United Kingdom
    Tk, read the whole thread before you make a comment :)
  19. Kursah

    Kursah

    Joined:
    Oct 15, 2006
    Messages:
    7,598 (2.68/day)
    Thanks Received:
    1,568
    Location:
    Missoula, MT, USA
    I'm interested to see further results...this isn't the first I've heard that the i7's may perform sub-par to current faster processors due to single-thread issues and such. With your access to this hardware, I'm hoping to see more results and some validation, you gave us your word, and for now that should suffice.

    To all those not interested or think this post is worthless...I don't feel it is, I think it could've initially been written and filled with screenshots and validations making a more solid point, but really it's not something we haven't already heard about and been warned of before. This is not new, but seeing some better proof of it could surely be interesting...but with future games hopefully supporting multi-core's more and more I'm sure the i7's will pull ahead...wonder how a 4 core i7 does against a quad core2 in Supreme Commander...only game I can think of that can utilize up to 4 cores..

    This thread has hope, and I look forward to seeing it grow positively...no need for harshness imo, if ya don't believe it or dislike the results...what's the point of increasing post-count for negative comments that aren't helpful? Sure the results could be bogus, guess we'll have to wait till next weekend, I for one will be watching and waiting.

    :toast:
  20. Darknova

    Darknova

    Joined:
    Nov 8, 2006
    Messages:
    5,037 (1.80/day)
    Thanks Received:
    535
    Location:
    Manchester, United Kingdom
    Kursah, you speak a lot of sense my friend :)

    Everyone else needs to stop ruining this for those who are interested in seeing the results. We have to wait until next weekend to see what happens, stop being so impatient, and stop with the inane comments.
  21. SimFreak47

    SimFreak47 New Member

    Joined:
    Sep 14, 2008
    Messages:
    557 (0.26/day)
    Thanks Received:
    37
    Location:
    In a city, thats in a state which is in the US, an
    imo a good comparison would be:

    use the core2 with x GB of RAM in Dual Channel
    use the i7 with x GB of RAM in Tripple Channel

    this would be benching the CPU's with their potential. ;)

    They should also be clocked at the same GHz (3.2GHz would be efficent)

    For testing games, you should have the same kind of hard drive. using a SSD and a HDD to bench games = fail. You need to have the same drive to get the most accurate results. IE, a 74GB Raptor.

    I'm not trying to sound like a rude prick. I'm just giving you a few suggestions ;)

    And when you get a chance, we'd all like to see some CPU-Z validation, with your TPU name under "Submitted by:"
  22. toloratedmeat New Member

    Joined:
    Oct 11, 2008
    Messages:
    34 (0.02/day)
    Thanks Received:
    6
    Yup. The screenies are comin right at cha. They will be up by Sunday.

    TPU name?
  23. FordGT90Concept

    FordGT90Concept "I go fast!1!11!1!"

    Joined:
    Oct 13, 2008
    Messages:
    13,319 (6.34/day)
    Thanks Received:
    3,344
    Location:
    IA, USA
    Um, what are you actually trying to prove? The best way to compare something is to follow the guidelines of the scientific method:
    1. Ask a Question
    2. Do Background Research
    3. Construct a Hypothesis
    4. Test Your Hypothesis by Doing an Experiment
    5. Analyzer Your Data and Draw a Conclusion
    6. Communicate Your Results

    At this point, I think we're stuck at step 1. Your post started out talking about the memory channel scheme (which is completely dependant on motherboard manufacturer implementation by the way) and ended with benchmarks comparing FPS. What exactly are you trying to prove? That triple channel is not as good as dual? That Nehalem's virtual-core approach is not as good as sticking to just physical cores?

    Science is only as good as the processes through which it is conducted. ;)


    Edit: It should be noted that games are not a very good source for processor benchmarks because a lot of it depends on the game design more so than anything else. The best benchmark for a processor is asymmetrical arithmetic and floating-point operations that engages all cores in non-repetitive work. Games, by nature, are symmetrical and as such, go no faster than the most burdened component (usually the graphics card). Because of this, most processor clocks are put to idle use and very little processor time is actually needed.
    Last edited: Oct 13, 2008
    Crunching for Team TPU
  24. Tatty_One

    Tatty_One Senior Moderator Staff Member

    Joined:
    Jan 18, 2006
    Messages:
    16,383 (5.29/day)
    Thanks Received:
    2,321
    Location:
    Worcestershire, UK
    Whatever the Op's intentions (and those of you who know me know I am always suspicsious of new members joining to post stuff like this, did they not recognise the merit of the forum before they had a message to spread?) :) But to be honest, there is some useful stuff here, and if he really does have the kit at his disposal then we may just get an early insight into the Pro's and Con's of Bloomfield, much of what he actually says is fairly logical, nevertheless I agree, he needs to be more consistent in his testing and provide more evidence but this may simply be down to over enthusiasm!

    I look foward to his further results.
  25. Wile E

    Wile E Power User

    Joined:
    Oct 1, 2006
    Messages:
    24,324 (8.56/day)
    Thanks Received:
    3,777
    All I want to know is, how far does it overclock, and on what voltages?

Currently Active Users Viewing This Thread: 1 (0 members and 1 guest)

Thread Status:
Not open for further replies.

Share This Page