hoping to start a debate :)

[parlee]

i hear people saying a lot lately that an opty can outpreform/overclock higher than any amdx2... my question is how come if the chips use the same core why dont they preform the same... people tell me of a stricter manufactoring process how optys are the best of the best x2's then why dont they sell for more rather than less on average?... debate plz 😀 (im not on either side so no shots at me plz 😛)

 

[RichPLS]

They perform nearly identical at stock, but overclock a bit higher making them perform faster o/c'd. Also run a bit cooler...

 

[parlee]

ive gathered that info by reading posts... but why do they oc better if its the same core?... is the stricter manufactoring process true? and if so why are they sold for cheaper on average...

 

[RichPLS]

Probably a marketing thing, big biz doesnt focus on initial cost as much as long term cost, and are not interested in 1xx opties. But pricing 939pin opties low competes and shows the techies that Opterons beat Intel Zeons as far as matching performance and killing them on costs.

 

[mpjesse]

Opteron's perform better because their on board memory controller is enabled. Since the processor is speaking directly w/ the memory there's less latency and overhead.

All Athlon 64/X2/FX/Opteron chips have onboard memory controllers. And technically yes, they all have identical cores (for each stepping/version). However only the FX and Operton have the memory controller enabled.

This is the reason why the FX and Opterons always own the X2's and Athlon 64's.

-mpjesse

 

[pat]

Opteron's perform better because their on board memory controller is enabled. Since the processor is speaking directly w/ the memory there's less latency and overhead.

All Athlon 64/X2/FX/Opteron chips have onboard memory controllers. And technically yes, they all have identical cores (for each stepping/version). However only the FX and Operton have the memory controller enabled.

This is the reason why the FX and Opterons always own the X2's and Athlon 64's.

-mpjesse

I don't understand.. what happen to the A64 memory controller then?? If it is disabled, and the chipset don't have one.. How can the memory be accessed?

 

[pat]

Opteron's perform better because their on board memory controller is enabled. Since the processor is speaking directly w/ the memory there's less latency and overhead.

All Athlon 64/X2/FX/Opteron chips have onboard memory controllers. And technically yes, they all have identical cores (for each stepping/version). However only the FX and Operton have the memory controller enabled.

This is the reason why the FX and Opterons always own the X2's and Athlon 64's.

-mpjesse

I think you mean that only FX and Opteron socket 940 has MULTIPLIERS unlocked.. A64, X2 and opteron has the multiplier locked..

Seriously, the memory controller has nothing to do with it ..

 

[Ned_Flanders]

...um, no. All A64's have an ODMC (On die memory controller). As Pat stated, you're getting the ODMC confuzzled with the limited multiplier selection of most A64's compared to the FX's.

 

[RichPLS]

Well, I hadware unlocked my Opteron (actually sent it to SoD for unlocking) and now run it at 23xFSB 250 😀

 

[parlee]

so by locking the memory controller AMD raises the price of amd x2's? if they had it unlocked from factory would they then be able to sell it cheaper than an opty?.. my understanding is that x2's are like fords, there mass production and the optys are the ferrari, they make much less but take more time on them, prove me wrong plz cause i just asked for it 😀

 

[mpjesse]

Taken from Anandtech.com:

You'll remember from our initial coverage that the major difference between the Athlon 64 and the Athlon 64 FX that the latter boasts a 128-bit memory controller as opposed to the 64-bit interface of the regular 64.

http://www.anandtech.com/cpuchipsets/showdoc.html?i=1941

Furthermore...

http://www.hardocp.com/article.html?art=NTI0

On the AthlonFX...

The packaging is a 940-pin lidded ceramic micro PGA. The lidded portion refers to the big metal heatspreader on top of the CPU, which is certainly a welcomed addition. This part runs at a 2.2GHz (11 x 200 CPU Clock) clock speed and supports Dual 64-bit DDR memory channels, which provides a 128-bit wide memory bus and 6.4GB/s of bandwidth working at DDR400 specifications. Fallback to DDR200 is supported.

On the Athlon 64...

The packaging is a 754-pin lidded organic micro PGA. Again, we see the heatspreader. This part runs at a 2.0GHz (10 x 200 CPU Clock) clock speed and supports an integrated single 64-bit DDR memory channel providing 3.2GB/s of bandwidth working up to DDR400 specifications. It also has fallback to DDR200 as well.

You're right, the memory controller is enabled on the Athlon 64 and X2. However, the memory bus is only 64 bits. The Opteron and FX have a 128bit memory bus. That makes a huge difference.

The memory controller has everything to do with it...

This article is obviously older but the memory bus stuff hasn't changed.

-mpjesse

 

[mpjesse]

http://www.hardocp.com/image.html?image=MTA2Mzk2MzI3NHFqaGJ6azR4emRfMV84X2wuZ2lm

Athlon64/X2 has one memory path (64bits) for RAM per core.

AthlonFX/Operton has two memory paths (128bits) for RAM per core.

So I was wrong... they both have working on die memory controllers. However there's a significant difference. Look at the image. 🙂

-mpjesse

 

[sturm]

socket 754 cores had single channel 64 bit on-die memory controlers.
Socket 939 cores have dual channel 64 bit on-die memory controlers for a 128-bit memory. This includes all socket 939 chips regular 64, x2, and fx.
Please get your info correct before you mislead people.

By the way, 2 of your links point to articles from january of 2004 and september of 2003. Its 2006 now!

Noticed your location "Location: Near a nuclear weapons lab" Did some of the nuclear material leak or are you part of the experiment?"


Sorry, boring day at work

 

[mpjesse]

I already stated those articles were old. Re-read my post.

Anyways my bad... I didn't know Socket 939 Athlon's had the 128bit memory interface. Now I know.

And for the record your mom was part of the nuclear experiments. She transfered some of the material to me when we had our affair. :lol:

"Your MOM goes to college!" -Napoleon Dynamite.

-mpjesse

 

[sturm]

Must be the cause of my problems also! :lol:

 

[mpjesse]

LOL

Seriously though, I live in Los Alamos, NM. The most educated city in America. Look it up. 🙂

-mpjesse

 

[Dealerluck]

According to your post you are overclocked 23x250 mhz. Even God doesnt have a computer running at 5.7 ghz.

 

[hazmat]

According to your post you are overclocked 23x250 mhz. Even God doesnt have a computer running at 5.7 ghz.

hehe owned.

Only a Pentium can come close to that. A few guys on ocforums.com overclocked a Pentium-D to 5.5 GHz with dry ice.

Im surprised that no one has said that another reason that the opterons OC better is because of the better quality chip. Microlithography is still an impeerfect science, sometimes theres errors in the fabrication, so they bin each chip according to its capabilities determined during testing at the factory. Low end chips with partly defective L2s end up as Semprons, and the top quality chips end up as Opterons, fill in the midrange with the "average" cpus and there you go.

 

[borandi]

According to your post you are overclocked 23x250 mhz. Even God doesnt have a computer running at 5.7 ghz.

hehe owned.

Only a Pentium can come close to that. A few guys on ocforums.com overclocked a Pentium-D to 5.5 GHz with dry ice.


THG have their video of a P4 to 5255 Mhz on Liquid Nitrogen :twisted: :twisted:

 

[Ned_Flanders]

Screw that (in your sig), join our Folding@home team

 

[borandi]

Screw that (in your sig), join our Folding@home team

FAH? PAH! Join BAH(boinc@hull)!

Its a project for my chemistry degree involving distributed computing. BOINC gives me more flexibility as I have to get local non-savvy people to participate. Also I prefer the wide ranging issues BOINC allows people to address and participate in. LHC is particularly cool. And Rosetta is the development project to make the program get protein structure more effectively than random shots in the dark with older versions (i.e. Predictor).

Then again, each to their own. Nicely.

20 computers strong 😀