We've been stiffed! Ram should be DDR2-667. Supplied=DDR2-533!!

We have been supplied with upgraded notebooks by our supplier. The Manufacturers recommended ram is DDR2-677 and the quote was for DD2-533 which was accepted and the supplier installed the quoted sub-par SODIMM before supplying the kit to us.

The ram issue was not picked up by our procuring department. Nor by the Test team who test all our OS/Workstations and parts like that.

Leaving aside the ethics of mis-selling sub-spec hardware, what long-term effects can we expect on our users equipment? Will it work ok, just a bit slower? Or can we expect lock-ups, BSODs and so on increasingly as the machines age?

Any websites/links with information we can use to batter the supplier with would be much appreciated.

 

What? You screwed up (your company) why do you ask about BSOD? Best Buy Has a Montevina with PC4200 listed I hope a typo not so sure now.

PC4200 and PC5300 makes little difference but well faster is better and all.

 

DDR2-667 is backward compatible with DDR2-533. You shouldn't run into any trouble. Depending on your use, the difference in speed might not even be noticed.

I don't believe there is any cause to argue with the supplier, as they sold you RAM that will work, as agreed upon.

 

You wouldn't notice any difference at all between speeds. The most important thing is the total size of ram.
For example: 1GB ram vs 4GB ram will do alot of difference but pc2-4200 vs pc2-6400 won't make any real difference.

But sure, they shouldn't supply you ram that isn't listed in the specs, though the user will never notice that, not even in 10 years as it will not cause any BSOD or such.

 

I wouldn't go that far. This is dependent on a number of factors including FSB speed and ram latency, so 4 CAS PC4200 would definitely be slower than 5 CAS PC6400. However the difference would not be highly noticeable to most users.

 

How exactly did you get stiffed if that's what your company bought?

It'll work fine, you won't have any issues running it.

 

No fault by the supplier there, they quoted you 533 and thats what you got. Also the manufacturers recommendation is just that, a recommendation. Not an absolute necesity.

 

I can bet anything that if i'd put 2 PC's infront of you with the same specs except the ram, one with CL4 533Mhz memory and one with CL5 800Mhz memory, that you won't notice any difference at all in performance, except in syntethic benchmarks, that's the only way to SEE the difference, but not feel it.

You know that CAS 4 is snappier than CAS 5? Lower cas latency is faster than higher. For example 667Mhz CAS4-4-4-12 vs 800Mhz CAS5-5-5-18 is evenly performance wise in syntethic benchmarks (means that 133Mhz memory speed can be compromised by lowering the CAS latency one snap). 667 vs 800Mhz with same cas has a difference of about 2% (to small amount to feel difference).
I'd personally preffer lower CAS than higher Mhz even though you won't even notice the difference in the first place.

Please tell me how memory is dependant on FSB performance wise? (Not compability wise). For example on a core 2 duo system the memory won't do that much performance wise, as the memory controller on a c2d-system isn't that great, as the memory calculations aren't done directly through the CPU.

I've written a big FAQ about memory and how it works togheter with motherboards/CPU on a swedish forum, so i wouldn't say that i don't know what i'm talking about here

 

That is why it is called latency. I discussed the difference extensively in the past here comparing CAS 4 pc4200 ram vs. CAS 5 PC5300 ram, where there ended up being no difference in actual speed. In the case of pc6400 ram vs PC4200, the increase in clock speed is sufficient to overcome the longer cycle latency. (e.g. ratios: 800/5=160, 533/4=133.25) As I said though, this would not be noticeable for most users.

There is no point in using ram clocked higher than the FSB as it will be downclocked to the FSB speed by the bios default FSB:ram ratio. This can be overcome through adjusting timings but definitely not for those with poor overclocking and bios tweaking skill sets.

I probably shouldn't have bothered saying anything because it was relevant more in an academic sense than a practical one, but being a technical kind of guy I like to chew on nuance and details. I am perilously close to being in over my head though.

 

Yeah, but you talked like you meant that cas4 should've been slower then cas5 in the first place

I would say it won't be noticeable in IRL performance at all, or would you?

No exactly what i meant. But you talked against me about this? That using ram specced at fsb isn't as good as memory running at higher speeds.. So what would you need 800Mhz memory for on a PC that won't need more than for ex 533Mhz ? As the 800Mhz memory will do 533 anyways on that system.

And memory will not always be downclocked according to FSB, it all depends on what motherboard and chipset you got.
My 945PM motherboard in my notebook doesn't do that, as i have a 5:8 dram:ratio original. So just because my cpu runs at 166Mhz(667Mhz quad) FSB doesn't mean that i need 667Mhz memory as lowest.
In my case i need 533Mhz memory and i can't change the memory divider nor anything that has to do with clock speeds.

Timings are easy to adjust on any PC with memset btw, works like a charm! So BIOS options for timings isn't always nessesary!

You can always chew on theory in all kinds of way, but the practical way of things are those things that means the most for the average users, being able to actually feel the difference or knowing why you won't feel and not only knowing that there is a very slight performance gain that you can never feel anyways.
Practical things are more relevant when it comes to this kind of thing IMHO, as there isn't any difference really (only in numbers).
I myself is a very detail-minded person too so i know what you mean.

 

All I said was that PC4200 ram with CAS 4 would be slower than PC6400 with CAS 5 - I don't think that implied higher latency meant faster performance, only that comparatively speaking the former ram would be slower than the latter.

I think you are confusing my posts with someone else. I never said anything of the kind.

Again, I didn't say that - I only stated that there would generally be no benefit to running ram at a frequency higher than the FSB, at least for the average computer user who knows little about this kind of thing.

I've never used this, I'll have to look into it. I don't tweak my notebook ram, and I when I ran my (now failed) crucial ballistix PC6400 ram on my desktop (Asus P5K) at CAS 4-4-4-12 and 900 MHz I found little (if any) difference in performance than at CAS 5-5-5-15 and 800 MHz. I totally agree, it probably only makes a significantly noticeable difference in benchmarks.

I hope that clarifies my thoughts a bit. This has become a lot more long-winded than I expected!

 

I would like to thank everyone for their contributions, it is obviously not as black and white as I believed....

 

You can use CPU-Z to check the timings of the RAM (the SPD page shows the available timing options while the memory page shows the timings being used).

The chipset is an important factor in memory performance. The new Intel chipset gives around 25% better bandwidth, even with DDR2-667 RAM. There's some results in part 2 of my Dell E6400 review. Also a single module will give reduced bandwidth and 2 modules the same size gives the best performance.

John