• Welcome to Overclockers Forums! Join us to reply in threads, receive reduced ads, and to customize your site experience!

Opteron low end. Why?

Overclockers is supported by our readers. When you click a link to make a purchase, we may earn a commission. Learn More.

Tyranos

Member
Joined
Jul 20, 2001
Location
OR
Why would anyone spend 250 dollars on an opteron 140 or 240, to have performance worse than even a dual mp 2500 system? Maybe I'm missing something here, but does anything else think that the low end opterons are almost a joke for the price? The core clock speed is so low, that its not even worth looking at I suppose. And, while the other model opteron 1XX's are quite a bit cheaper than their 2xx counterparts, the 140 is the same price or more than the 240 at lots of places. :\
 
Well, do you actually have benchmarks comparing the two systems? I haven't seen too many of those.
Second of all, some server stuff might benefit/need to use 64-bit applications, and the Operton is sure a lot cheaper than an Itanium 2.
 
--------

Despite my criticism, I'm very tempted to buy a 240 right now. So far, I've only seen a 400mhz overclock on an opteron with ln2 cooling. I wonder if its not entirely the cpu's fault. The current boards out now only support fsb changes, and no multiplier changes. So to get an opteron 240 at a respectable speed, one would have to raise the fsb to an insane level. The 240's multiplier should be around 8.5 or 9. Since no one has removed the heatspreader from an opteron that I know of, maybe it is possible to remove it and edit the bridges (if there are any on the surface) based on the technical docs from AMD. This would let us finally see what type of overclocking potential an opteron has, not limited by extreme front side busses. Of course we would need to put the heat spreader back on, or find other means of protection since the new cores are very fragile. What say fellow overclockers?
 
Last edited:
I don't think the problem of low overclocks was to do with the low multiplyer, I thought it was because of the onboard memory controller, maybe it isnt as good as the northbridges on the NF7-s for example?
 
They had the fsb to over what 220? Thats on par with a good nf7s. The lack of multiplier adjustment severely limited their overclocking ability. And what do you do after you reach a fsb limit? Multiplier time. I think that it may be the memory that was holding them back as well, but at that high of a fsb its hard to tell. Vcore on the SK8N has already been modified for higher values.

I'm also wondering if they have even tried non registered higher spec memory with those chips! It doesn't "support" them, but neither does an MP system. BUT I've seen non registered stuff work in dual mp computers. Please inform me if I'm not, but I'm thinking of many different possibilities that no one has touched upon yet. (as far as I know)
 
Last edited:
Some guys tested the opteron with ln2 and reached only 2.25ghz. At amdzone they managed the same with aircooling, which would (as mentioned) show that the bottleneck is not the cpu itself, rather the memory controller or something similar.
 
Tyranos... The main selling point of the Opterons is the 64bit CPU
structure along with backwards compatibility of all 32bit applications. Perception of this particular cpu not performing as well against any Intell P4 3GHz (+/-) is irrellavant under 32bit applications. AMD is pinning all bets on 64bit appls coming out in the next couple of years, therefore migration from 32bit over to 64bit software/OS would be without any sort of hickups by IT professionals. Secondly, Opterons will only run on mobos with registered mem modules as of recent documentation/ recommendations from AMD itself. To wit, there has not been any Opteron specific mobos out now without the registered memory slots... The thought behind this, of cource is system integrety. Server/workstation for professional/government offices would benefit from a configuration built with reliability and redundancy.

User level Opeterons, which would bring up the Athlon64, would then be the ones for destop level users. Whether it's for the enthusiust or OEM builders, this particular CPU would become the leader for destop user. Non-registered mem modules would then apply for these cpus.

The expensive in producing the Opterons for mainstream users would not benefit AMD at all. Therefore, Athlon64 CPU are what AMD thinks and believes mainstream users would purchase and make the money AMD is banking on. Producing A64's at 2.0GHz and above is what AMD is targeting for... Operterons came out at the lower speed due to the fabrication/manufacturing process that AMD had to deal with. Therefore we are seeing Opterons below the 2GHz range. Their performance test levels not being up to par with Intel P4 is only a perception, realworld testing and hands on usage with these cpu will tell the difference.
 
There is like 20-40 no connect pins on the processor so there is a possibility those might be multiplier pins. But there was like no mention of a multiplier in the datasheet, so I think mb makers can't even put in a mechanism to adjust the multiplier becasue AMD doesn't say it can be even accesed or programed.
 
Dual cpu desktops for more average users are becoming more common. As far as I know, no Athlon 64 will be smp capable. So, the people who want a dual processor system for everyday use will be forced to look at opterons. Whether or not the dual desktop market is large enough for AMD to consider is questionable though.

I've also seen results from opterons running in 64bit mode in the linux 64b os and windows xp 64 bit edition beta. They become much more efficient, but not as a huge leap as one would come to think.
 
I agree that the 140s and such aren't very cost-effective, but the strength of the Opterons is their scalability - Let's say a 146 is on par with a 3.0Ghz P4. Then a 2-way 246 Opteron would handily beat out the 2-way 3.06Ghz Xeons in almost every test, and especially ones that rely on lots of memory access. Let's not even talk about the XeonMPs.

It is difficult to overclock Opteron systems because the HT busses aren't happy running above spec - 800Mhz is already a lot, and they aren't out-of-sync like AGP and PCI are on some boards today.
 
slow down guys. opteron's aren't even mainstream yet. who's making chipsets for these CPU's other then AMD? wait awhile. let AMD make their money in the enterprise sector and when things trickle down to us we'll have much better chipsets and better stepping CPU's with better memory controllers.

it's the story of the computer. hurry up and wait.
 
Here's mine

fb3cbcdb.jpg


fb3cbcfd.jpg


fb3cbd0f.jpg


fb3cbcc0.jpg


fb3bbfd1.jpg


managed to get it up to 1.8GHz at default Vcore 1.55v

fb3b3275.jpg


Got to get a pair of PC2700 ECC REG to push up the mem bench, at the moment running PC2100

9800 Pro stock / 1.8GHz
 
Interesting. Does the asus have bus locks? Or is it pushing agp/pci out of spec? You score about 1200 points above my best. Though I have a 9700pro and 210fsb.
 
Why do the futuremark entries have 200mhz as the fsb? 200x8 != 1800. Strange 3dmark.
 
Opteron 244 performance is not exactly stunning by a long shot.
Check here and see how it loses most of benchmarks to Pentium 3.2C and yet it costs more. Motherboards cost more as well, and registered memory costs more too.
The only reason to buy that would be to work with 64-bit applications. But that will not take over any time soon and when it does the CPU will be outdated anyway.
Opteron 140, 142 are ultimate jokes but don't you dare to question wisdom of AMD, these people are magicians. They managed to survive while making financial losses for years, and they manage to make financial losses even when they had good and competetive processors. All around a magic company.

Did they sell many Opterons? No, but that is not the point, what they are doing is making a lot of hype, delays, rumours, and so on. Then maybe, just maybe they'll make some money, mostly by selling to the same crowd that spreads these rumours at the first place. I surely won't buy any.
 
This isn't a thread to bash AMD or opterons as a whole. Don't post those types of things in my thread. And they ARE selling many opterons. Who bought 10000 for their new supercomputer? Who just bought 2000 for their new supercomputer? What part of the market are you looking at anyway, Somalia's? AMD has garnered many clients for these chips. Amazing the 1.8ghz chip loses against the 3.2ghz chip. Why don't you try a 2 or 2.2ghz opteron and see what the outcome is? Comparing the absolute top end Intel 32 bit desktop cpu to AMD's mid range, first batch 64 bit server chips and then pointing out like its a monstrous feat that the Intel came out on top somewhat? Give me a break. The reason the top end opterons cost so much is because they are new ENTERPRISE hardware. The people who buy them are businesses. I don't think AMD really cares if YOU can buy one, because the corporations are the ones with money that WILL pay those prices. This thread was to question the lowest range Opterons because those don't offer anything new performance wise other than 64bit. They do, however beat out other chips at the same speeds.
 
whatever2003 said:
.

Did they sell many Opterons? No, but that is not the point, what they are doing is making a lot of hype, delays, rumours, and so on. Then maybe, just maybe they'll make some money, mostly by selling to the same crowd that spreads these rumours at the first place. I surely won't buy any.

Oh so all those super computer builders dont know what they are doing using opterons?:rolleyes: Are they selling many? - a damn sight more than Intel are selling Itaniums!
 
Back