Does a high FSB really make a difference?

[GRagland]

I think im going to get a 2500+ Barton or 1700+ 1.5v, but why should i get $100 pc3200 when i can get $50 pc2700? Does having a fsb of 200 instead of 166 really make much of a difference? I have the feeling that it doesnt matter that much and it is not worth the money to get expensive ram. Could someone do a bench to test the performance difference?

 

[Jeff7181]

In a word, yes.

I would normall expand on that, but I've been sitting here too long. Someone else can explain it, or post some benchmark scores that prove it. If they haven't by the time I check the boards again, I'll post some stuff for ya.

 

[pspada]

No, a high FSB makes little difference.......unless you are looking for performance. These days, anything less than PC3200 and you will regret it, I guarantee.

 

[GRagland]

yeah, thats what everyone says, but it would be nice to see some benches to back that up.

 

[Gage8]

Getting 3200 over 2700 will not make a noticeable difference unless you OC. I have Corsair XMS 2700 that is rated for 2 3 3 7 at 166FSB (333/2=166). However, if I try to OC it I have to slow the CAS down to 2.5.

The point is that I'm going to buy some 3200 so i can gain more stablility at higher FSB speeds which gives me a overall faster system. However, 3200 would not do a thing for me if I kept the FSB at 166 which is what 2700 is rated for.

 

[Harabecw]

just look at the fact P4 2.4's with 800FSB are as fast as P4 2.6 533FSB's. look in anand's own reviews...
sometimes people forget the forums have a hardware review site with em

 

[ChampionAtTufshop]

for waht its worth
default 3dmark2001se benchmark @133mhz fsb = 12thousandish
default 3dmark2001se benchmark @200mhz fsb = 16thousandish

vidcard was kept at default the entire time, only thing changed was fsb

almost all sites that review mobo's end up ocing ab it and some give you performnace of oc/not oc so search the web

 

[Jeff7181]

Here's some results... anandtech deletes posts too quick, the ones I had posted before that had some extensive RAM testing I did are gone already.

Excel Document

 

[TerryMathews]

I can back it up mathematically.

The formula to calculate RAM bandwidth is as follows:
(Speed in MHz) * (8)

The 8 is due to the RAM width in bits (64) converted to bytes (8)

That formula needs to be * (2) if you're using a DDR system.

Then you multiply the sum by the efficiency. This is usually ~80% for a system running CAS 2.

So, for our examples, a 166MHz FSB would yield ~2100MB/s of RAM bandwidth. An identical system running a 200MHz FSB would yield ~2500MB/s. Most applications and games are starved for RAM bandwidth, not CPU horsepower.

 

[Jeff7181]

I proved it in some benchmarks I ran in REAL games, not synthetic benchmarks... but like I said, anandtech deleted that post already... if I get bored tomorrow maybe I'll do it again.

 

[TerryMathews]

Originally posted by: Jeff7181
I proved it in some benchmarks I ran in REAL games, not synthetic benchmarks... but like I said, anandtech deleted that post already... if I get bored tomorrow maybe I'll do it again.

My number isn't a benchmark. It's how much bandwidth is there. Fact. If you run Sandra, you'll get a lower number becuase sandra isn't getting exclusive use of the RAM bandwidth.

It's like taking a car to a dyno. You get an approximation of what the horsepower is. I just blueprinted the engine and can tell you what the exact performance is at that given point.

 

[Jeff7181]

Originally posted by: TerryMathews

Originally posted by: Jeff7181
I proved it in some benchmarks I ran in REAL games, not synthetic benchmarks... but like I said, anandtech deleted that post already... if I get bored tomorrow maybe I'll do it again.

My number isn't a benchmark. It's how much bandwidth is there. Fact. If you run Sandra, you'll get a lower number becuase sandra isn't getting exclusive use of the RAM bandwidth.

It's like taking a car to a dyno. You get an approximation of what the horsepower is. I just blueprinted the engine and can tell you what the exact performance is at that given point.

Bad example with the car... you'll never EVER be able to tell what the EXACT performance is at ANY given point. Dyno reading are more accurate than someone's math done based on engine specs. You can determine the absolute maximum power based on the airflow through the heads... but engines are not 100% efficient, so that won't be accurate either. Some people even argue a dyno doesn't even tell you the EXACT numbers... but I won't get into that... WHOLE different subject for a WHOLE different message board.

Anyhow... the benchmarks I did were real world performance in real games. Not many people trust synthetic benchmarks to give an accurate depiction of performance anymore.

 

[pillage2001]

It's all about the bandwidth. It's how fast data can be pumped to the RAM or any other peripherals.

 

[TerryMathews]

Originally posted by: Jeff7181
Anyhow... the benchmarks I did were real world performance in real games. Not many people trust synthetic benchmarks to give an accurate depiction of performance anymore.

Right, but that's not a synthetic benchmark. Big difference.

 

[Jeff7181]

Originally posted by: TerryMathews

Originally posted by: Jeff7181
Anyhow... the benchmarks I did were real world performance in real games. Not many people trust synthetic benchmarks to give an accurate depiction of performance anymore.

Right, but that's not a synthetic benchmark. Big difference.

May not be a synthetic benchmark... sure you can say more is better, and 2.5 GB/s is more than 2.1 GB/s... but what does that tell me about what it's actually going to do when I play a game? Not a whole lot... and even less to a novice.

 

[GRagland]

So let me get this straight:

No matter what chip I am getting (even if it is a bad oc'er), it would be benificial to get pc3200 because i can drop the multiplier and raise the fsb. And that would give me a substancial performace boost worth the extra money, without even actaully overclocking (mhz of the chips stays the same).

So the advantages of pc3200 over pc2100 is NOT that pc3200 it is better for overclocking, cause any ram will let you overclock the same distance mhz wise, but that the advantage is that once you REACH YOUR MAX OVERCLOCK, then you can adjust the fsb/multiplier ratio to achieve better performance. (and of couse better performance= higher fsb)

 

[TerryMathews]

Originally posted by: Jeff7181

Originally posted by: TerryMathews

Originally posted by: Jeff7181
Anyhow... the benchmarks I did were real world performance in real games. Not many people trust synthetic benchmarks to give an accurate depiction of performance anymore.

Right, but that's not a synthetic benchmark. Big difference.

May not be a synthetic benchmark... sure you can say more is better, and 2.5 GB/s is more than 2.1 GB/s... but what does that tell me about what it's actually going to do when I play a game? Not a whole lot... and even less to a novice.

By you're logic, people should question CPU freqs *in the same processor model* as well. They don't. People automatically assume more is better, which is correct. More is better. End of story. There is no down side to having more RAM bandwidth.

 

[TerryMathews]

Originally posted by: GRagland
So let me get this straight:

No matter what chip I am getting (even if it is a bad oc'er), it would be benificial to get pc3200 because i can drop the multiplier and raise the fsb. And that would give me a substancial performace boost worth the extra money, without even actaully overclocking (mhz of the chips stays the same).

So the advantages of pc3200 over pc2100 is NOT that pc3200 it is better for overclocking, cause any ram will let you overclock the same distance mhz wise, but that the advantage is that once you REACH YOUR MAX OVERCLOCK, then you can adjust the fsb/multiplier ratio to achieve better performance. (and of couse better performance= higher fsb)

You hit the nail on the head. Get the processor overclocked as much as you can so that it's still stable, then drop the mult and increase the FSB so that it works out to the same clock speed. Free (well, not exactly free) performance.

PC3200 is RAM that is warranted to work at 200MHz FSB, so it takes one variable out of the overclocking equation. You can overclock some RAM, but most PC2100 won't run that high. I've heard of some PC2700 hitting it though.

It's all about your budget. If you don't have $100+ to allot for RAM, then go with cheaper stuff and hope for the best. But more expensive, higher quality RAM will let you get higher FSBs (within reason, don't expect to hit 233 easily) without too much trouble.

 

[GRagland]

Thank you! But im still interested in maybe a ut2k3 benchmark comparing cpus at the same mhz speed, but different ratios of multi/fsb. (all using the same ram timings). Cause that would help me determine if its worth the dough. Know of any website that has a comparison like this?

 

[DAPUNISHER]

Originally posted by: GRagland
Thank you! But im still interested in maybe a ut2k3 benchmark comparing cpus at the same mhz speed, but different ratios of multi/fsb. (all using the same ram timings). Cause that would help me determine if its worth the dough. Know of any website that has a comparison like this?

Read this He doesn't often increase FSB without clockspeed in tow but if you do a search for aThugsrook you may find that paricular eval. He's done extensive benchies many times with different boards, ram, CPU's.

 

[GRagland]

Thugsrock posted this:

P4 12x200fsb 2.4ghz = 12417 3DMarks
P4 18x133fsb 2.4ghz = 12001 3DMarks

Machu posted this:

13x170fsb 2.2ghz = 17041 3DMarks
10x220fsb 2.2ghz = 17828 3DMarks


Hmm... thats not much of a performance boost for the extra amount of money that ddr400 costs.
Ill keep looking, see if i can find a real world benchmark.

 

[TerryMathews]

Originally posted by: GRagland
Thugsrock posted this:

P4 12x200fsb 2.4ghz = 12417 3DMarks
P4 18x133fsb 2.4ghz = 12001 3DMarks

Machu posted this:

13x170fsb 2.2ghz = 17041 3DMarks
10x220fsb 2.2ghz = 17828 3DMarks


Hmm... thants not much of a performance boost for the extra amount of money that ddr400 costs.
Ill keep looking, see if i can find a real world benchmark.

Ummm, $50/400 3dmarks = 8 3dmarks per dollar. That's a great price/performance ratio. I wish I could put an extra $50 in my comp and gain 400 3dmarks.

Gimme a minute and I'll post a 3dmark score of a system I built for a friend recently. >200MHz FSB.

EDIT: He pegged 12,950 on an AXP at 1850MHz with a GF4Ti 4200 @ 380/580 and RAM at CAS 2.5 205MHz

 

[GRagland]

Okay, I'm getting a 2500+ and Epox-8RDA+, what ram would you recommend. 2x256 or 1x512 is what i want. I was thinking about Winbond CH-5 2x256 PC3200 which i can get for $94 shipped. But, I heard that the 2500+ has been having trouble at very high fsb speeds, so it may be pointless to get such good ram. Here's a review of the Winbond CH-5.

Also, since i have a 40gb and a 80gb (8mb cache) hard drive, would it be worth it performance wise to get a abit nf7 2.0 which has RAID ($40 more then the 8RDA+)?

 

[Jeff7181]

Originally posted by: TerryMathews

Originally posted by: Jeff7181

Originally posted by: TerryMathews

Originally posted by: Jeff7181
Anyhow... the benchmarks I did were real world performance in real games. Not many people trust synthetic benchmarks to give an accurate depiction of performance anymore.

Right, but that's not a synthetic benchmark. Big difference.

May not be a synthetic benchmark... sure you can say more is better, and 2.5 GB/s is more than 2.1 GB/s... but what does that tell me about what it's actually going to do when I play a game? Not a whole lot... and even less to a novice.

By you're logic, people should question CPU freqs *in the same processor model* as well. They don't. People automatically assume more is better, which is correct. More is better. End of story. There is no down side to having more RAM bandwidth.

So then tell me, exactly how many FPS will I gain in UT2k3 if I had a P4 2.4 and switched to a P4 3.0?
You can't... and that's my point... REAL WORLD RESULTS are more useful than all the specs in the world.

 

[mngisdood]

IIRC the higher bandwidths are more benefitial to P4 processors (esp. p4c) than to AXP's. I'll try to dig up some numbers...