What is the OC difference in the Core 2 Duos?

When Overclocked, what is the difference between the Core 2 Duos?

e4300
e4400
e6300
e6320
e6400
e6420
e6600
e6700

All I know is that with decent aftermarket cooling and with a GA965PDS3, the e4300 can be OCd to 3ghz.

Does the extra cache make a difference? I have heard that it makes no difference.

What are the different levels that these CPUs can typically be OCd to with a P965, with a 975X, with a 650iSLI, with a 680iSLI, with a P35???

Is the price difference between the e4300 and the e6600 worth it? What kind of OC can be expected from these CPUs?
19 answers Last reply
More about what difference core duos
  1. Good questions! The tricky part is that the CPUs are moving targets. Even if it's the same part number, it might be at least a different stepping from what someone else has used, and of course even chips from the same lot vary. Not surprisingly, Intel gets better at minimizing the OC margin available as it gains more production experience with a chip design, so the best overclockers are often the early versions of a given design.
    Also, since overall CPU performance is much more dependent on the core frequency than on the FSB frequency, given two CPUs with the same rated core freq, but different FSB freqs, the CPU with the *lower* nominal FSB can often be overclocked to a higher performance level. Some useful questions to get an idea of a CPUs overclockability potential:
    1) Do "sister" CPUs (CPUs of the same design) exist that run at higher FSBs freqs?
    2) Do "sister" CPUs exist that run at higher core freqs?
    3) Do "sister" CPUs exist that have higher thermal output?
    "Yes" answers are good, because they imply that the underlying CPU design can support a higher value.
  2. Quote:
    Good questions! The tricky part is that the CPUs are moving targets. Even if it's the same part number, it might be at least a different stepping from what someone else has used, and of course even chips from the same lot vary. Not surprisingly, Intel gets better at minimizing the OC margin available as it gains more production experience with a chip design, so the best overclockers are often the early versions of a given design.
    Also, since overall CPU performance is much more dependent on the core frequency than on the FSB frequency, given two CPUs with the same rated core freq, but different FSB freqs, the CPU with the *lower* nominal FSB can often be overclocked to a higher performance level. Some useful questions to get an idea of a CPUs overclockability potential:
    1) Do "sister" CPUs (CPUs of the same design) exist that run at higher FSBs freqs?
    2) Do "sister" CPUs exist that run at higher core freqs?
    3) Do "sister" CPUs exist that have higher thermal output?
    "Yes" answers are good, because they imply that the underlying CPU design can support a higher value.


    whatever you just said, I think I knew. Anyway its hard to tell.

    What is the advantage of getting any Core 2 Duo over an e4300. Is the average OC higher? I know that you can have two processors with the same name that OC very differently, but then the whole idea of buying a cpu is a gamble.

    is it that an e4300 can be overclocked most often to 3ghz, an e6300 to 3.1, an e6400 to 3.2, and an e6600 to 3.4?

    btw im talking about 100% stable OCs not benchmarkers.
  3. Quote:
    ...

    is it that an e4300 can be overclocked most often to 3ghz, an e6300 to 3.1, an e6400 to 3.2, and an e6600 to 3.4?...

    Just about. Within a line with the same FSB, the answer should be "yes". Between lines (as with the e4300 vs e6300) it's a bit trickier. For example, the e4300 might be able to go to a higher core freq than the e6300 if what's holding the e6300 back at the limit is the circuitry involved with the FSB, not the core. For example, OCing an e4300 by 25% results in a 2.25GHz core and 1000MHz FSB. OCing an e6300 by 25% results in a 2.265GHz core and 1333MHz FSB. Although both CPUs can probably work fine with a ~2.25GHz core speed, it's much more likely that the e4300 will work OK with a 1000MHz FSB than that the e6300 will work OK with a 1333MHz FSB. If the e6xxx series could normally reach 1333MHz FSB speeds, Intel most probably would have already released 1333MHz FSB CPUs.

    One of the big unknowns here is *why* a given e4300 chip is sold as an e4300 rather than as an e6300? Is it because it failed a 1066MHz FSB test? If not, perhaps Intel just needed chips to sell as e4300s for marketing strategy reasons. Even if it did fail such a test, did it fail by much? A slight voltage boost might "rescue" it at the cost of a modest lifetime reduction that is acceptable to the OCer, if not to Intel.
  4. Quote:
    ...

    is it that an e4300 can be overclocked most often to 3ghz, an e6300 to 3.1, an e6400 to 3.2, and an e6600 to 3.4?...

    Just about. Within a line with the same FSB, the answer should be "yes". Between lines (as with the e4300 vs e6300) it's a bit trickier. For example, the e4300 might be able to go to a higher core freq than the e6300 if what's holding the e6300 back at the limit is the circuitry involved with the FSB, not the core. For example, OCing an e4300 by 25% results in a 2.25GHz core and 1000MHz FSB. OCing an e6300 by 25% results in a 2.265GHz core and 1333MHz FSB. Although both CPUs can probably work fine with a ~2.25GHz core speed, it's much more likely that the e4300 will work OK with a 1000MHz FSB than that the e6300 will work OK with a 1333MHz FSB. If the e6xxx series could normally reach 1333MHz FSB speeds, Intel most probably would have already released 1333MHz FSB CPUs.

    One of the big unknowns here is *why* a given e4300 chip is sold as an e4300 rather than as an e6300? Is it because it failed a 1066MHz FSB test? If not, perhaps Intel just needed chips to sell as e4300s for marketing strategy reasons. Even if it did fail such a test, did it fail by much? A slight voltage boost might "rescue" it at the cost of a modest lifetime reduction that is acceptable to the OCer, if not to Intel.

    that last part is binning right?

    So why would anyone get anything above an e4300?
  5. Yep. People buy higher-rated chips for the guarantee that it will work at the higher speed under standard conditions. We don't know how many e4300s could have been e6300s, and how many didn't pass the testing for an e6300.

    One clue: if there is a shortage of high-end chips of a given design, the lower-rated chips probably can't be OC'd that much. For example, since AMD can't even match the performance of the upper end of Intel's C2D line, it's sure not selling lots of CPUs at lower speed grades than they bin to.
  6. Quote:
    Yep. People buy higher-rated chips for the guarantee that it will work at the higher speed under standard conditions. We don't know how many e4300s could have been e6300s, and how many didn't pass the testing for an e6300.

    One clue: if there is a shortage of high-end chips of a given design, the lower-rated chips probably can't be OC'd that much. For example, since AMD can't even match the performance of the upper end of Intel's C2D line, it's sure not selling lots of CPUs at lower speed grades than they bin to.


    So it is worth it to get an e6600 over an e4300 because it is guaranteed that it will run at 2.4ghz with standard conditions. This means that the chip has a higher chance than the average e4300 to run at 3ghz. Then again, getting a e4300 that was binned because they needed more e4300s, but can perform like an x6800, is just getting lucky. Anyhow, there is a higher chance of getting an e6600 that was binned too low, than getting an e4300 that can run at that speed. (i.e. when they put a high performing cpu in a lower bin, they pick the next lower bin first)

    Also, what is this about steppings? I seem to remember reading that stepping g is good, but cannot remember anything else. is the stepping printed on the cpu? How can you tell before buying what stepping you will get?
  7. Quote:
    Not surprisingly, Intel gets better at minimizing the OC margin available as it gains more production experience with a chip design, so the best overclockers are often the early versions of a given design.


    I doubt if that statement even have any truth in it. From my experience, the better Intel get in making the same family of chips, the better they will overclock. Also, as time progress, Intel will even reduce FSB on higher end chip and rebrand them as lower end chip to mass sell them.

    I am using an i975Xa_YDG. My first CPU running came from my laptop after I upgraded its CPU, which was a T2300. That thing will not run any faster than 2.3MHZ, from the original 1.66 Ghz. Recently, I was fortunately enough to get my hand on a T2250 ( which is really a T2600 in disguise with 13X multipliers. Intel rebrand most T2600s into T2250s when they introduce the Merom family to mobile computing). And with the T2250, I am now humming along @ 3.030 Ghz ( 13X233FSB). Alot of ES or early T2600 couldn't even get any where near 2.8ghz. I am now eyeing the T5200 and T5300 ( Meroms with reduced FSBs, they look tasty)
  8. OCing is pure luck. If you can get a e4300 to 3.2 Ghz, it doesn't mean I can with mine. Most of the time, lower FSBs will yield the most % of OC.
  9. Quote:
    OCing is pure luck. If you can get a e4300 to 3.2 Ghz, it doesn't mean I can with mine. Most of the time, lower FSBs will yield the most % of OC.


    yea but a 50% OC on an e4300 is 2.7ghz and a 40% OC on an e6600 is 3.36ghz. If the chance is roughly similar, then the e6600 is good, right?
  10. Quote:
    Good questions! The tricky part is that the CPUs are moving targets. Even if it's the same part number, it might be at least a different stepping from what someone else has used, and of course even chips from the same lot vary. Not surprisingly, Intel gets better at minimizing the OC margin available as it gains more production experience with a chip design, so the best overclockers are often the early versions of a given design.
    Also, since overall CPU performance is much more dependent on the core frequency than on the FSB frequency, given two CPUs with the same rated core freq, but different FSB freqs, the CPU with the *lower* nominal FSB can often be overclocked to a higher performance level. Some useful questions to get an idea of a CPUs overclockability potential:
    1) Do "sister" CPUs (CPUs of the same design) exist that run at higher FSBs freqs?
    2) Do "sister" CPUs exist that run at higher core freqs?
    3) Do "sister" CPUs exist that have higher thermal output?
    "Yes" answers are good, because they imply that the underlying CPU design can support a higher value.


    whatever you just said, I think I knew. Anyway its hard to tell.

    What is the advantage of getting any Core 2 Duo over an e4300. Is the average OC higher? I know that you can have two processors with the same name that OC very differently, but then the whole idea of buying a cpu is a gamble.

    is it that an e4300 can be overclocked most often to 3ghz, an e6300 to 3.1, an e6400 to 3.2, and an e6600 to 3.4?

    btw im talking about 100% stable OCs not benchmarkers.

    I think FSB and cache can make a somewhat difference. I have oced 4400 to 2.4 which is the same frequency as 6600. I am using two similar setups with same kind of memory. But FSB is different. 6600 has native 1066 . When i crunch numbers with super pi,for example, 6600 can run faster than 4400 oced at 2.4. So imo , fsb and cache make difference inaddition to mere frequency.
  11. Quote:
    Good questions! The tricky part is that the CPUs are moving targets. Even if it's the same part number, it might be at least a different stepping from what someone else has used, and of course even chips from the same lot vary. Not surprisingly, Intel gets better at minimizing the OC margin available as it gains more production experience with a chip design, so the best overclockers are often the early versions of a given design.
    Also, since overall CPU performance is much more dependent on the core frequency than on the FSB frequency, given two CPUs with the same rated core freq, but different FSB freqs, the CPU with the *lower* nominal FSB can often be overclocked to a higher performance level. Some useful questions to get an idea of a CPUs overclockability potential:
    1) Do "sister" CPUs (CPUs of the same design) exist that run at higher FSBs freqs?
    2) Do "sister" CPUs exist that run at higher core freqs?
    3) Do "sister" CPUs exist that have higher thermal output?
    "Yes" answers are good, because they imply that the underlying CPU design can support a higher value.


    whatever you just said, I think I knew. Anyway its hard to tell.

    What is the advantage of getting any Core 2 Duo over an e4300. Is the average OC higher? I know that you can have two processors with the same name that OC very differently, but then the whole idea of buying a cpu is a gamble.

    is it that an e4300 can be overclocked most often to 3ghz, an e6300 to 3.1, an e6400 to 3.2, and an e6600 to 3.4?

    btw im talking about 100% stable OCs not benchmarkers.

    I think FSB and cache can make a somewhat difference. I have oced 4400 to 2.4 which is the same frequency as 6600. I am using two similar setups with same kind of memory. But FSB is different. 6600 has native 1066 . When i crunch numbers with super pi,for example, 6600 can run faster than 4400 oced at 2.4. So imo , fsb and cache make difference inaddition to mere frequency.

    how much of a difference, 3seconds?
  12. Quote:
    Good questions! The tricky part is that the CPUs are moving targets. Even if it's the same part number, it might be at least a different stepping from what someone else has used, and of course even chips from the same lot vary. Not surprisingly, Intel gets better at minimizing the OC margin available as it gains more production experience with a chip design, so the best overclockers are often the early versions of a given design.
    Also, since overall CPU performance is much more dependent on the core frequency than on the FSB frequency, given two CPUs with the same rated core freq, but different FSB freqs, the CPU with the *lower* nominal FSB can often be overclocked to a higher performance level. Some useful questions to get an idea of a CPUs overclockability potential:
    1) Do "sister" CPUs (CPUs of the same design) exist that run at higher FSBs freqs?
    2) Do "sister" CPUs exist that run at higher core freqs?
    3) Do "sister" CPUs exist that have higher thermal output?
    "Yes" answers are good, because they imply that the underlying CPU design can support a higher value.


    whatever you just said, I think I knew. Anyway its hard to tell.

    What is the advantage of getting any Core 2 Duo over an e4300. Is the average OC higher? I know that you can have two processors with the same name that OC very differently, but then the whole idea of buying a cpu is a gamble.

    is it that an e4300 can be overclocked most often to 3ghz, an e6300 to 3.1, an e6400 to 3.2, and an e6600 to 3.4?

    btw im talking about 100% stable OCs not benchmarkers.

    I think FSB and cache can make a somewhat difference. I have oced 4400 to 2.4 which is the same frequency as 6600. I am using two similar setups with same kind of memory. But FSB is different. 6600 has native 1066 . When i crunch numbers with super pi,for example, 6600 can run faster than 4400 oced at 2.4. So imo , fsb and cache make difference inaddition to mere frequency.

    how much of a difference, 3seconds?

    4 seconds
  13. Quote:

    One of the big unknowns here is *why* a given e4300 chip is sold as an e4300 rather than as an e6300? Is it because it failed a 1066MHz FSB test? If not, perhaps Intel just needed chips to sell as e4300s for marketing strategy reasons. Even if it did fail such a test, did it fail by much? A slight voltage boost might "rescue" it at the cost of a modest lifetime reduction that is acceptable to the OCer, if not to Intel.


    Your partly wrong, the E4x00 series is a different design to the E6x00 series as the E4x00 chips have a full 2meg cache (true Allendale) and the die is smaller than the E6x00 series chips (Conroe) which has a full 4meg cache, apart from the E6300 and E6400 that had 2meg cache, in which the half of the 4meg cache is turned off(some were named Allendale and some were named as true Conroe's) which possibly had a failure in the cache area to keep yields high.

    Any failures within the chip(core and/or cache) would be rebranded as Celeron-L, a single core version of the Core2 Duo with 1 core and parts of the cache turned off leaving 0.5meg, and also as Pentium E21x0 series with half the E4x00 cache disabled.

    As Intel is selling full versions of E4x00(2meg) and E6x00(4meg) with everything fully working, you'd be wondering whats happened to the failures? It looks like Intel's getting all the failed stock and rework them into Celeron's and Pentium E21x0's which is due for an imminent release in the next few weeks.
  14. Okay, to set things straight.

    There are only 3 stepping's of Intel processors that you can buy now. The B2 stepping of Conroe which was the launch stepping. Kentsfield was released as the B3 stepping and finally the new L2 which covers the Allendale Core.

    This information can be seen below in this chart which is found at this link: Processor Finder

    sSpec# CPU Speed Processor # Bus Speed Mfg Tech Stepping Cache Size Package Type
    SL9ZF 2.66 GHz E6700 1066 MHz 65 nm B2 4 MB LGA775
    SL9S7 2.66 GHz E6700 1066 MHz 65 nm B2 4 MB LGA775
    SL9ZL 2.40 GHz E6600 1066 MHz 65 nm B2 4 MB LGA775
    SL9S8 2.40 GHz E6600 1066 MHz 65 nm B2 4 MB LGA775
    SL9T9 2.13 GHz E6400 1066 MHz 65 nm L2 2 MB LGA775
    SL9S9 2.13 GHz E6400 1066 MHz 65 nm B2 2 MB LGA775
    SLA4T 2.13 GHz E6420 1066 MHz 65 nm B2 4 MB LGA775
    SLA3F 2 GHz E4400 800 MHz 65 nm L2 2 MB LGA775
    SL9SA 1.86 GHz E6300 1066 MHz 65 nm B2 2 MB LGA775
    SLA4U 1.86 GHz E6320 1066 MHz 65 nm B2 4 MB LGA775
    SL9TA 1.86 GHz E6300 1066 MHz 65 nm L2 2 MB LGA775
    SL9TB 1.80 GHz E4300 800 MHz 65 nm L2 2 MB LGA775


    The next stepping for Conroe will be the G1 or G2. These will be the the new E6x50 processors with 1333. The next Allendale core is supposed to be either M0 or M1. Of course until Intel launches these new parts we won't know for sure what stepping they will actually be. :?

    The Allendale Core is supposed to support the upcoming Pentium Core line of processors E2140, E2160 processors and the new Celeron 440 and 460 processors. The Pentium line will have 2 cores and 1Meg cache and the new Celeron will have a single core and 512M of cache.
  15. Quote:
    Okay, to set things straight.

    There are only 3 stepping's of Intel processors that you can buy now. The B2 stepping of Conroe which was the launch stepping. Kentsfield was released as the B3 stepping and finally the new L2 which covers the Allendale Core.

    This information can be seen below in this chart which is found at this link: Processor Finder

    sSpec# CPU Speed Processor # Bus Speed Mfg Tech Stepping Cache Size Package Type
    SL9ZF 2.66 GHz E6700 1066 MHz 65 nm B2 4 MB LGA775
    SL9S7 2.66 GHz E6700 1066 MHz 65 nm B2 4 MB LGA775
    SL9ZL 2.40 GHz E6600 1066 MHz 65 nm B2 4 MB LGA775
    SL9S8 2.40 GHz E6600 1066 MHz 65 nm B2 4 MB LGA775
    SL9T9 2.13 GHz E6400 1066 MHz 65 nm L2 2 MB LGA775
    SL9S9 2.13 GHz E6400 1066 MHz 65 nm B2 2 MB LGA775
    SLA4T 2.13 GHz E6420 1066 MHz 65 nm B2 4 MB LGA775
    SLA3F 2 GHz E4400 800 MHz 65 nm L2 2 MB LGA775
    SL9SA 1.86 GHz E6300 1066 MHz 65 nm B2 2 MB LGA775
    SLA4U 1.86 GHz E6320 1066 MHz 65 nm B2 4 MB LGA775
    SL9TA 1.86 GHz E6300 1066 MHz 65 nm L2 2 MB LGA775
    SL9TB 1.80 GHz E4300 800 MHz 65 nm L2 2 MB LGA775


    The next stepping for Conroe will be the G1 or G2. These will be the the new E6x50 processors with 1333. The next Allendale core is supposed to be either M0 or M1. Of course until Intel launches these new parts we won't know for sure what stepping they will actually be. :?

    The Allendale Core is supposed to support the upcoming Pentium Core line of processors E2140, E2160 processors and the new Celeron 440 and 460 processors. The Pentium line will have 2 cores and 1Meg cache and the new Celeron will have a single core and 512M of cache.


    So, when you buy an e6300 or e6400, you can get either an Allendale or Conroe core? How does this affect OC capability?

    How does one know what stepping he will be getting by buying a CPU apart from reading the label on the cpu itself?
  16. mb, perhaps I got too carried away with extending my e4xxx vs e6xxx example. However, it does seem that there are e4300s and e6300s with the same stepping, as well as versions with different steppings.

    My point is that sometimes a given chip has really failed a binning test for a faster-rated CPU, sometimes it hasn't but is being used because of chip supply issues, and sometimes it hasn't been tested for a faster rating. The first case is worst for OCers, the 2nd case is best, and you never know which of the three cases you'll get!
  17. Quote:
    Not surprisingly, Intel gets better at minimizing the OC margin available as it gains more production experience with a chip design, so the best overclockers are often the early versions of a given design.


    I doubt if that statement even have any truth in it. From my experience, the better Intel get in making the same family of chips, the better they will overclock. Also, as time progress, Intel will even reduce FSB on higher end chip and rebrand them as lower end chip to mass sell them....)
    We're both right. I'm talking about the early stages of a design's life cycle. You're talking about the later stages.

    At the very beginning, Intel is likely to be conservative on rating chips, thus allowing more OC potential (in a decent design). As they gain more process/design experience, they tend to introduce more speed grades and apply looser binning standards=less OC headroom (as they gain more confidence in the predictions of the binning tests). Later, as you describe, the manufacturing process (hopefully) gets so good that they have more than enough CPUs in the higher grades and can tighten binning standards, dropping some of the better-performing CPUs down into lower bins (for more OC headroom on average).
  18. Quote:
    ...
    How does one know what stepping he will be getting by buying a CPU apart from reading the label on the cpu itself?

    You can only tell by looking at the sspec number on the packaging or CPU.
  19. The only advise I can give you is that there seems to be more reports of E6xxx processors reaching higher overclocks than the E4xxx processors. And like other members have already said. It comes down to a crap shoot. If the market is really going after the lower clock processors and actual binning is pushing more parts out at a higher clock then to meet demand Intel will just down bin them and us enthusiasts get to report great overclocking on the lower end parts. :D

    Mondoman: I will have to disagree with you on the binning process. For each and every processor that Intel sales they must meet a minimum guardband. The guardband is the margin from specifications that Intel sets for their processors ie(Temperature mostly, Vcc of the processor, Frequency) What this means is that if Intel can't meet the necessary guardband then they can't sell them at a higher frequency and make more money. We are currently seeing this for the QX6800 processors. These are constrained and what I hear if Intel could make more they could sell them for those ridiculus ~$1250 price tags.

    Take AMD's fastest X2 processor 6000+. It is at the very limit of its guardband. That is one of the reasons why it can't overclock well. They have pushed the 90nm process and K8 architecture about as far as it can go. The other problem for AMD is their new 65nm process is still not up to the same speed as the older 90nm process. :(

    So to summarize. Intel is always looking to bin out all of the die on every wafer at as High a Frequency as possible. During the validation and development of a new manufacturing process they must reach a certain number of processors binned out across the marketing speed range they plann to sell at. As time goes on the number of fast parts that are binned out increases over time. They also come out with new steppings that fix bugs, speed path issues and minimizes hot spots on die to again allow more higher biined parts.
Ask a new question

Read More

CPUs Core