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

Is ocz EB really worth it over pc4000 in a a64 setup?(paging hitech)

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

Thecalmingapple

Member
Joined
May 29, 2004
Im purchasing my system tomorrow night and seriously need ram advice as there seems to be so many different opinions on a64 memory.

Hitech's a64 post states that if buying new ram ddr500 is wanted even at higher latencys. Then in almost every other post people swear up and down that eb is the only way to go for an a64. Some people say tight timings are important even if you have to use a divider, and others say that higher speeds at loose timings out performs it. To tell you the truth, im confused about all of this and seriously need a answer soon. Any help is appreciated.
 
everyone keeps telling me whenver i post that my ram rules the gods for a64.


i think it is the Enhanced bandwith thing that makes the difference, that, plus its overclockability with micron chips.

what you really want to look at is what chips overclock best on an nforce 3 setup for a64. the speed doesnt matter neccessarily. im sure hitech means that you should grab some pc400 speed rated micron -5bc
 
seems like intel chipsets and the 64 systems tolerate cas 3 better that the nf7 boards. they like cas2.5/2. a good set of the eb pc3700 can run at 3.2.3.5 at about 2.8volts(anandtech) most of the pc4000+ is designed to run at 2.5.4.4.7 or 3.4.4.8.(although some of the newer stuff can run a little better timings) so i guess the answer is how fast are u gonna be able to run the fsb and cpu.. if it will only run 250/260 then u may want to consider the lower latency eb series from ocz.(or pc4000- not as good timings) if u have the capability to run like 250 and can set the ratio to 5/6, and u want a crazy memory oc, (if the mb can handle it.) u may want to get some of the ocz el pc4400. it is primo ram, and has been known to run 300+ mhz

if this is confusing, it just that there are alot of varilables that go into matching ram with ur setup.. goodluck flapper
 
Well the multi will be 10 so if running at 1:1 250 with tight timings would be the best bet right?
 
Thecalmingapple said:
Well the multi will be 10 so if running at 1:1 250 with tight timings would be the best bet right?

if u think the ur rig is capable of 250 1-1,the eb sounds like a good match for u
 
What are the variables that would make me not capable to run at those speeds? mobo voltage and such?

EDIT: btw, I will be running a A64 mobile 3200+
 
Last edited:
i dont know alot about the a64 characteristics, but first of all u have to have a good idea of how fast the cpu will go (fsb) secondly how well will the mb run?? will it run over 250? 3rd is the ratio.. will it run well 1-1 or maybe even 5/6 like maxvala' s a64 ran. i wouldnt get ANY ram till i knew these basics first..
 
But how am I supposed to know the max fsb of my chip without having memory to run it to find out?
 
Hello, everyone, this is my first post. I have some OCZ PC 3500 EB on my Athlon 64 system. (In case you wanted some quick specs: MSI K8N NEO PLATINUM, Athlon 64 3200+, 512 MB OCZ PC 3500 EB, ATI RADEON 9800 PRO). I'm not super impressed with it but then again, I have no problems with it. I'm not really into overclocking that much but I have pushed to a little bit over 230. My RAM had no problems with it, whatsoever. Then again, that is really not that big of an overclock at all. If you can get it cheap, you might as well go with it but I definitely would not pay more for the EB series.
 
The pc4000 Im comparing the OCZ Eb to does:
2-3-2 DDR400+, Some have reported 2-2-2@200-220. <all of [email protected] or less
2.5-3-3-5/2.5-4-3-5 @ DDR500+ @ 2.75v
3-3-3@PC4400 @ 2.85v

It seems to do pretty well, and thats why Im questioning spending the money on EB if I could get memory like this for much much cheaper.
 
Well im not really worried about price, im just wondering which one will perform better in an a64 setup. Because if the performance of the ocz is worth the extra price, by all means I will spend the extra cash. But I was just wondering because hitechs guide reccomends to buy pc4000 even at higher latencys(which I really trust), yet everyone else reccomends eb.
 
The EB does not do that much for me. It is not like you are going to be seeing some amazing performance or anything with it. It is just like any other high - end RAM .
 
Thecalmingapple said:
Well im not really worried about price, im just wondering which one will perform better in an a64 setup. Because if the performance of the ocz is worth the extra price, by all means I will spend the extra cash. But I was just wondering because hitechs guide reccomends to buy pc4000 even at higher latencys(which I really trust), yet everyone else reccomends eb.

The DDR500 (PC4000) I refer to does not conflict with the OCZ EB, which can run 250+ MHz at low equivalent latency.

I brought up DDR500 as A64 opens up memory bus to 250-300 MHz, whereas 250 MHz has been the max for Nforce2 (taking into account of high Vdimm, memory overclocking). One should try to take advantage of such high memory speed. On the other hand, there is tradeoff between tight timing x-2-2-2 with relax timing at high speed, also there is tradeoff between 1T and 2T. I have done some estimates on these tradeoff, but premature to quote them at this point.

From
A64 CPUs, chipsets, motherboards

Originally posted by hitechjb1

For A64,
- Can run any speed (ASYNC) using the CPU_memory_divider, to match memory module speed.
- Slow or faster memory can be used to get whatever bandwidth allowed in the memory modules.
- ASYNC has minimum impact on memory bandwidth efficiency
- 754 efficiency 95%+, 939 efficiency 86%+

...

Memory modules (for 754 and 939 platforms)

Socket 940 Opteron and FX platforms require the use of registered memory modules, in ECC mode for mission-critical systems.

A nice thing about 754 and 939 CPU's is that they support, for non-mission critical system, existing unbuffered, DDR 400/500 (PC 3200/4000) memory modules that most of us have.
Both 754 and 939 support either ECC or non-ECC memory modules.
754 supports up to 3 dimms. 939 supports up to 4 dimms.

PC3200 or DDR 400 is rated at 200 MHz, max bandwidth = 200 x 2 x 8 = 3200 MB/s
PC3500 or DDR 438 is rated at 219 MHz, max bandwidth = 219 x 2 x 8 = 3500 MB/s
PC3700 or DDR 462 is rated at 231 MHz, max bandwidth = 231 x 2 x 8 = 3700 MB/s
PC4000 or DDR 500 is rated at 250 MHz, max bandwidth = 250 x 2 x 8 = 4000 MB/s
PC4400 or DDR 550 is rated at 275 MHz, max bandwidth = 275 x 2 x 8 = 4400 MB/s

For A64 system, the memory bus frequency can be flexibly set at various (SYNC/ASYNC) ratio to CPU frequency and FSB setting (see post on overclocking setting) with minimal impact on memory bandwidth efficiency.

Existing DDR400 (PC3200) modules would work fine with A64 754 and 939 system.

If getting new memory modules, DDR 500 memory module, or even DDR 550 are preferred, even with higher tRCD and CAS latencies. With the DDR500+ (PC4000+) memory modules, memory bus can be overclocked to the level 250 - 300+ MHz. Memory bus at such 250 - 300+ MHz level delivers significantly higher raw memory bandwidth than that at the low 200 - 230 MHz even with lower latancy of x-2-2-2 by most PC3200 modules. Though some existing programs, drivers may not be able to take full advantage of the high raw bandwidth, have to see whether future programs, OS can benefit such excess bandwidth.

Some low latency modules with DRAM chips such as BH5 (new memory modules with BH5 are hard to get currently (2004)) may be able to overclock to 230-250 MHz with tight timing x-2-2-2 and enough high Vdimm (if choosing to use 3.0 - 3.3 V).

There are modules with
- Micron 256Mb DRAM chips (-5BC), found in some Buffalo PC3700, OCZ EB
- Hynix chips (BT-D43), found in some Kingston PC3200, to the 250-275+ MHz with 2.x V
- HyperX PC4000 (SamSung TCCC chips) rated 2.6 V to at least 250 MHz
- Samsung TCCD chips (4ns) to 250 MHz level with CAS 2.5
- A-DATA Vitesta modules to 250-300 MHz
Running at 2.x V would lessen some concerns about high Vdimm on the CPU memory controller interface.

OCZ EB series, e.g. PC3700, overclock very well to high bus frequency (DDR550 speed) with low equivalent latency.

There are tradeoffs in overall performance (not just raw memory bandwidth) between low latency x-2-2-2 at 230-250 MHz with 1T (command rate), and higher latency between x-3-3-2.5 and x-4-4-3 at 270-300 MHz with 2T (command rate), test out these two cases for specific applications and benchmarks.

For 939 system, as the dual channel 128-bit memory bus can already provide 80+% bandwidth over the 754, a tight x-2-2-2 running 240 - 250 MHz may provide sufficient memory bandwidth for the whole system using overclocking memory such as with BH-5 chips.

I've seen 939 system achieving 90%+ memory bandwidth efficiency.
Assuming 90% bandwidth efficiency, with good memory modules for high bus frequency (probably at relaxed timing),
at 250 MHz, the effective bandwidth would be 7200 MB/s,
at 275 MHz, the effective bandwidth would be 7920 MB/s,
at 300 MHz, the effective bandwidth would be 8640 MB/s (doable for 4400 module).

Based on my reading of the Anandtech 939 run-up test using Mushkin PC3500 Level 2 and OCZ PC 3500 Platinum (both use BH5 DRAM chips), with 2 dimms (not clear 256MBx2 or 512MBx2), at 200 MHz memory bus, 10-2-2-2 1T timing, the Sandra (SP2) memory banwdith test obtained was 6100 MB/s (95.3% efficiency) for MSI K8N neo2, 6000 MB/s (93.8% efficiency) for ABIT AV8 and became 5000 MB/s (78.1% efficiency) for 2T.

For a typical 754 system, the effective memory bandwidth is around 3000-4000 MB/s.


Towards year end of 2005, A64 platforms may move to DDR2 memory module (discussed in other posts in this thread) which would require new motherboards, maybe new CPU socket (not clear as of May 04).

From:
Memory bus, cache and memory bandwidth (for 940, 754, 939)
Memory bus frequency setting, SYNC/ASYNC mode
Memory modules (for 754 and 939 platforms) (post 14)

This is just my opinion, comments are welcome.
 
Another suggestion, memory modules are expensive, a big % of the system cost.

Why not use existing modules to test out the whole system first, and then add expensive memory later with thorough testing. Then you know what you are getting and what is improving.
 
flapperhead said:
i dont know alot about the a64 characteristics, but first of all u have to have a good idea of how fast the cpu will go (fsb) secondly how well will the mb run?? will it run over 250? 3rd is the ratio.. will it run well 1-1 or maybe even 5/6 like maxvala' s a64 ran. i wouldnt get ANY ram till i knew these basics first..

Agree.

For A64, apart from 1:1 ratio, the memory setting and memory_divider depend on FSB, CPU multiplier, ..., and one has to take into account the limit of CPU, FSB, memory, .... as illustrated in this post.

hitechjb1 said:
The impact of running non 1:1 memory:FSB ratio on memory bandwidth efficiency has been discussed and established few months back, and found to have minimum or no impact on memory efficiency. This is also illustrated by d]g[ts's results in the two links he listed, attaining 97% efficiency in both cases.

The question between a higher multiplier CPU (e.g. 3200+ NC CG with x11) or a lower multiplier CPU (e.g. 3200+ NC CG with x10) is reduced to the question of having enough memory divider to locate the sweet spot for a given CPU and memory module(s) used.

E.g.
NC 3200+ with x11
available memory divider = 14, 12, 11, 10, 9, ... (through x11, x10, x9, ... CPU_multiplier in combination of 1:1, 5:6 memory:FSB ratio)

NC 3000+ with x10
available memory divider = 12, 11, 10, 9, ... (through x10, x9, ... CPU_multiplier in combination of 1:1, 5:6 memory:FSB ratio)

So the difference between NC 3200+ and NC 3000+ is only the 14 memory divider.

memory_bus_frequency = CPU_frequency / memory_divider

Unless the memory modules used is very slow (< CPU_freq_max/12), the 14 memory divider is usually immaterial for air/water overclocking, hence NC 3000+ would be OK. Extreme cooling may be different though, to be discussed if needed.

This link shows how to compute the memory divider for a given CPU multiplier, memory:FSB ratio.

Memory bus frequency setting, SYNC/ASYNC mode (scroll down a page)


hitechjb1 said:
Given the 3700 EB can run high memory bus (250+ MHz) with low equivalent latency, for air/water cooling of CPU, a 3000+ NC CG with 10x is what I would get. Saving ~$50 than a 3200+ for other things.

In principle, a 2800+ may suffice, but given the price difference between 2800+ and 3000+ is not much (after price drop), a 3000+ is preferred.
 
hitechjb1 said:
Another suggestion, memory modules are expensive, a big % of the system cost.

Why not use existing modules to test out the whole system first, and then add expensive memory later with thorough testing. Then you know what you are getting and what is improving.

The only memory I have is pc100. Unless I could test the boundaries of my system with that, or find pc3200 that will dirt cheap, the extra cost of buying ram, and sending back I don't think will be worth the trouble.

The Mobile athlon 64 3200+ I will be purchasing has a 10x multiplier. Now assuming I can run at a 1:1 ratio, 250 fsb will be all I need to run at the what seems to be maximum of 2.5ghz of these chips. But others say that 1:1 doesnt effect performance at all, so 250 would not be needed as I could use a divider and get the same performance.

What do you think I should do?
Buy memory with tighter timings, use a divider, and use my highest multi?
buy Eb and run 250 with loose timings at 1:1, with my highest multi?
buy the pc4000 that does :

2-3-2 [email protected] or less
2.5-3-3-5/2.5-4-3-5 @ [email protected]
3-3-3@PC4400@ 2.85v

Like I said, im really confused what to buy..but based on your opinions it seems like the pc4000 would be a better choice to test out my systems potential, and maybe buy different memory later based on the results?
 
Back