Very interesting 1700+ vs. 2500+ Results

[high]

Here's the lowdown:

This isn't a review and I am lazy, but I did a few 1700+ vs 2500+, each at 2.3ghz, and have gotten some weird results.

Rig # 1 - Gaming Rig -
2500+ @ 2.3ghz ( 10.5x219)
512 OCZ 3200 EL (11 3 2 2)
Radeon 9800np (420/340)
80 WD 8MB HDD
DFI Lan Party NF2 Ultra

Rig # 2 - HT PC -
1700+ @ 2.3ghz (11.5x200)
512 Generic 2100 (11 2 2 2.5)
Geforce 2 MX 400 (5700 Ultra tomorrow)
120 Maxtor 8MB HDD
MSI K7N2 Delta - L

Test 1 - PC Mark -

Rig #1:
CPU: 6687
RAM: 6325
HDD: 1154

Rig #2:
CPU: 6908
RAM: 3970
HDD: 1458

Test 2 - SiSoft '04
Combined Performance Index Wizard

Rig #1:
Arithmetic : 11946
MultiMedia : 41026

Rig #2:
Arithmetic: 12304
Multimedia: 42820

Now I know these arent the best benches and I'm still chugging away. But the question is "why is my HT PC (backup and usually crappier pc) beating my big bad barton with a higher fsb in my gaming rig?"
I am just wondering if people could post their scores to have more than 2 rigs to compare, and if we can get to the bottom of the 1700/1800 vs 2500 debate.

Thanks and I appreciate any imput.

 

[myocardia]

Hah, why do you think I still have my old XP2400 plugging away? It will only do 205mhz fsb max, but when I run it at 12x200, it's plenty fast enough for me. Although, I should say that all of those PCMock benchmarks are much more dependent on overall clock speed than to amount of level 2 cache, or system fsb. Why not try doing a few benches with something like SuperPi, or maybe the SETI benchmark app?

 

[Supa]

Maybe it's the hard drives. #2 HDD score is higher, both 7200rpm?


---

 

[CaiNaM]

doesnt make sense to me.. i just replaced a 1700+ w/ a 2500+ in my wife's machine yesterday, and the performance difference is day and night.

also, is that 5700u (tomorrow) something you're buying? if it is, consider a 5900nu or 5900se - much better performance and same price range ($160-170).

 

[AkumaX]

cas 2 vs cas 3?

 

[Corey0808]

I don't think that you can compare the two tests. The reason being is the fact that the setups are totally different. If you just switched processors I think these tests would be more accurate. Just my opinion

 

[adams828]

Originally posted by: Corey0808
I don't think that you can compare the two tests. The reason being is the fact that the setups are totally different. If you just switched processors I think these tests would be more accurate. Just my opinion

i'd have to agree on that. you've got 2 different mobos, 2 different memory sticks (and timings!).. so you can't really call it a CPU benchmark, more of a entire system benchmark.

so it that case:
for your own use, if the 1700+ "feels" faster (and benches faster), then why not make it your primary machine?

 

[high]

Thanks for the awesome input guys....here are my arguments.
I am running the 1700+ at 3:2 with 2100 RAM....11 2 2 2.5...on a mobo that should not be outperforming my DFI Lan Party, (was 2.5x the cost). And both mobo's are nforce2 Ultra chipsets. Also, the 2500+ has a lower multi a lot higher FSB running 1:1 with 3200 OCZ. You would think that it would DOMINATE the 1700+...but I ran Seti @ Home, Super PI, and ALL the results favour the 1700+. And you're right Adams828, about the gaming rig, I'm just tryin to justify why I put it in my media center (crappier pc) and the hdd and proc are benching higher than my gaming rigs parts.

 

[high]

Also, Cainam: I'm in Canada.....5900SE's run for $270 or so Canadian, which actually, now that I'm typing this, isn't too bad compared to 225 for the 5700 Ultra, but keep in mind it won't be a gaming rig, more of a media pc. Will I get better clarity with a 5900 vs 5700 on my 20" TV, Epson projector?

 

[sugarkang]

you have something misproperly configured.
your numbers just aren't possible in a correctly configured rig.

here's a thought:
just plug your 1700+ chip into your 2500+ rig and see what numbers you get.

my guess is that you've got the ram running ASYNC with the cpu.
big penalty there. that's exactly what happened to me when i upgraded from 1700 to 2500 and ultimately OCed to 3200. once i got the async fixed, it was all gravy.

 

[high]

my ram in my media center is 3:2, so yes it's async. My gaming rig is 1:1 @ 219, 1.775 vcore 2.8 vdimm.... heh I'm trying to find reasons why I shouldn't swap chips, very lazy But seriously. I just dont know why that chip is dummying this 2500+ - extra 256 l2, higher fsb, lower multi, 1:1 with ram....etc

 

[adams828]

there must be some kind of setting/tweak somewhere that is throwing this off. i noticed, the PC mark memory scores *are* much higher for your gaming rig (as they should theoretically be). so the confusing thing is, what would cause the 2500+ to score lower on the PC mark CPU test than the 1700+ did ?

uhm.. no idea. update bios? i'll try and think harder once i get some work done lol

 

[DAPUNISHER]

How many times did you run each bench? SiSuck and PCMock also don't do a good job of showing overall real world system performance either, as you are already aware of. You need to run each bench at least 3 times with clean boots in-between each run then find the mean of the scores to have anything close to accurate benchmarking results, otherwise it's just poor testing methodology.

Now, if you play a game or do something else that can appreciate the extra system bandwidth and/or extra cache space you'll see the advantage decidedly goes to the Barton in the nF2. However, I think you'll find that with the 1700+ in the nF2 with all other things being equal, that for most tasks the extra cache doesn't provide the significant improvement the way AMD's PR rating suggests it does@the same clockspeed. The only real difference I noticed between my Barton 2500+@2.43GHZ and t-bred B 2100+@2.4GHZ was the snappier feel of the system overall that others also claimed was their experience. Beyond that I just felt since they seemed about equal otherwise when something I was doing did benefit from the extra cache, I'd have it

 

[CaiNaM]

Originally posted by: high
Also, Cainam: I'm in Canada.....5900SE's run for $270 or so Canadian, which actually, now that I'm typing this, isn't too bad compared to 225 for the 5700 Ultra, but keep in mind it won't be a gaming rig, more of a media pc. Will I get better clarity with a 5900 vs 5700 on my 20" TV, Epson projector?

heh. dunno.. don't use the vivo components on my card ;p

i feel for ya tho; here in the states the 5900nu w/ vivo is the same price as a 5700u, which is why i suggested that in the first place.