CPUZ 1.73 Benchmark Thread

[soccerballtux]

Based on the information that has been posted here, it appears that two things are happening:

1). Everyone is seeing MT results that are (at best) equal to n-1/n * 100% of what they should be seeing. For example, someone with a quad is seeing 75% of the MT result that they should get for the chip
2). Anyone using two logical cores is seeing LESS performance than 50% since the benchmark is loading up an extra background thread on one core and doing practically nothing with the second core

Should be interesting to see what the fix does to address the problem. Then I'll know if my estimates are correct.

btw, I reran it at 4.7 GHz: 1141/3332

http://www.xtremesystems.org/forums/attachment.php?attachmentid=132437&d=1440227194

The memory for my first run was DDR3-1600 CL9 (better than I thought, oh well)
The memory for my 4.7 GHz run was DDR3-2400 CL10

might be intentional to make a well rounded bench. a cpu's ability to juggle two threads at once on the same core is also indicative of performance

 

[Burpo]

Looks like your record is safe YBS1.. I didn't hit the Silicon Lottery.. Chip stumbles above 200bclk.. working on it.. Volts gettin uncomfortable.

 

[jpiniero]

Notice how for the pentium III the ST and MT scores are the exactly the same, which they would be if Bench wasn't borked.
Intel's pride and joy, the Atom heralded the era of Fast Enough computing and affordable Ultra Small Form Factor notebooks, though it was quickly slammed by critics for its inability to fluidly playback HD flash video (let's say, of cats) inside a browser.

Funny, I ran it on a Coppermine Celeron Ghz and got 74. I thought it would be the lowest in this thread.

 

[YBS1]

Looks like your record is safe YBS1.. I didn't hit the Silicon Lottery.. Chip stumbles above 200bclk.. working on it.. Volts gettin uncomfortable.

I'm relieved and disappointed at the same time. When I first saw your screenshot I saw the mythical "number" and went "Weeeeee! Someone on here finally did a benchable 5!" After my post I realized the way we (or our mainboards) do our overclocks you were at 4.8 on 6 core loads.

So is it still a better chip than the one you've been using? I mean as far as daily overclock goes?

 

[Burpo]

Barely.. I can't crack 12.. Best so far was at 197 bclk and hit 11.93. After 197 I get "A clock interrupt was not received on a second processor within allocated time".

 

[MongGrel]

$80 Xeon X5670

Nice

Wish I had a Rampage, but can't gripe about the P6T7 much.

 

[know of fence]

Funny, I ran it on a Coppermine Celeron Ghz and got 74. I thought it would be the lowest in this thread.

You also were on to something historically.
Does not having SSE instructions, mean the CPU can't run this bench at all or will it perhaps run it slowly with MMX or whatever?

The next generation Celeron was the 'Coppermine-128' (sometimes known as the Celeron II). These were a derivative of Intel's Coppermine Pentium III and were released on 29 March 2000. Like the Mendocino, the Celeron-128 used 128 KB of on-chip L2 cache and was (initially) restricted to a 66 MHz Front Side Bus Speed, But the big news was the addition of SSE instructions, due to the new Coppermine core.

 

[soccerballtux]

$80 Xeon X5670

can you help me do this? I've decided I want to stick with a DDR3 system for the forseeable future

socket bracket to adapt? what motherboard?

 

[richierich1212]

can you help me do this? I've decided I want to stick with a DDR3 system for the forseeable future

socket bracket to adapt? what motherboard?

Good luck trying to find a X58 mobo for a decent price.

 

[jpiniero]

You also were on to something historically.
Does not having SSE instructions, mean the CPU can't run this bench at all or will it perhaps run it slowly with MMX or whatever?

It does have SSE (1).

 

[MongGrel]

can you help me do this? I've decided I want to stick with a DDR3 system for the forseeable future

socket bracket to adapt? what motherboard?

Good luck trying to find a X58 mobo for a decent price.

I think Burpo has one of these, but not positive.

http://www.amazon.com/Asus-Rampage-III-CrossFireX-Motherboard/dp/B003H2AQG6

Yeah, scoring a Mobo cheap for a X58 can be a bit of watching these days, I picked up an ASUS P6T awhile back for about $130 that I have a X5650 in for the HTPC.

If you do not all ready have an older X58 board, might not be worth doing.

Still use one of these old guys on the main myself, but I picked it up cheap open box off the Egg many years ago.

http://www.amazon.com/SOCKET-PROCES...UTF8&qid=1440278518&sr=1-1&keywords=ASUS+P6T7

 

[SPBHM]

It does have SSE (1).

Katmai P3 added SSE as far as I know (slot 1, with external l2 cache) before Coppermine,

also, I think some posts earlier it was mentioned that the 32bit version uses x87!?

http://forums.anandtech.com/showpost.php?p=37645739&postcount=193

 

[BonzaiDuck]

When I saw this thread, showing that the new version 1.73 had both "bench" and "stress" built in, I decided to give it a try.

I'll have to study the additional comments such as those by Burpo and MongGrel, to appreciate the nuances.

I used the i7-4790K as reference for my 2700K @ 4.7Ghz.

I should've posted the Bench results with the raw scores, rather than the percentages. Like I said -- I have to read the posts some more.

 

[Burpo]

Just don't check "Reference"

@soccerballtux.. As has been said, finding an X58 board that overclocks well is the thing. Ones w/SATA3 & USB3 get expensive.
This is the deal today..
http://www.ebay.com/itm/Genuine-Del...-7591-H869M-/181837722079?hash=item2a565f71df

That and an $80 Xeon with good HSF & decent memory is all you need.

 

[BonzaiDuck]

Just don't check "Reference"

Yup. I'd run it both ways, but did the screen capture with the box checked.

I'm assuming that my own results point up the "future-proof" nature of the 5-year-old SB K processors. That is, "no upgrade hurries." Just looking at these scores, given my deferral for building a newer system, it just looks like paraphrase from "Treasure of the Sierra Madre:" "Devils Canyon? We don' need no Devils Canyon! What do we need stinkeen Devils Canyons for?"

Of course, it would be interesting to see how an OC'd 4790K compares to the "reference" 4790K. Maybe someone posted just such results.

LOOKING BACK AT POSTS: This seems to satisfy my need for more info-- confirming that the reference was not OC'd:

My Second System.

CPU: Intel Corei7 4790K Without Overclock.

Why does my rating down Guys.

 

[alcoholbob]

Yup. I'd run it both ways, but did the screen capture with the box checked.

I'm assuming that my own results point up the "future-proof" nature of the 5-year-old SB K processors. That is, "no upgrade hurries." Just looking at these scores, given my deferral for building a newer system, it just looks like paraphrase from "Treasure of the Sierra Madre:" "Devils Canyon? We don' need no Devils Canyon! What do we need stinkeen Devils Canyons for?"

Of course, it would be interesting to see how an OC'd 4790K compares to the "reference" 4790K. Maybe someone posted just such results.

LOOKING BACK AT POSTS: This seems to satisfy my need for more info-- confirming that the reference was not OC'd:

You'd have to compare two sets of data with the same memory timings/speed as well since the single threaded benchmark can vary by over 100 points at the same clock speed but with different memory timings.

 

[soccerballtux]

setting aside the power usage building one really makes sense given the lackluster 14nm showing, 10nm delays, and 6nm wall we're about to hit. If Zen is 40% better IPC I might upgrade but...there would be no reason to leave my 32GB RAM. When I need 64 (in 3 years?) just pick some up cheap off ebay and be good for another 5.

 

[know of fence]

Katmai P3 added SSE as far as I know (slot 1, with external l2 cache) before Coppermine,

also, I think some posts earlier it was mentioned that the 32bit version uses x87!?

http://forums.anandtech.com/showpost.php?p=37645739&postcount=193

You are totally right, Pentium 3 despite being one of the first CPUs to have SSE instuctions, is only availabe as a reference for the 32 bit benchmark.
To hit a lower score than Pentium 3, I could probably take a Pentium 75 MHz (16 MB RAM) and run this bench in windows 95/98, though I'm not sure if CPU-Z supports windows 98.
Much more interesting would be to see the lowest x64 score.

 

[SPBHM]

You are totally right, Pentium 3 despite being one of the first CPUs to have SSE instuctions, is only availabe as a reference for the 32 bit benchmark.
To hit a lower score than Pentium 3, I could probably take a Pentium 75 MHz (16 MB RAM) and run this bench in windows 95/98, though I'm not sure if CPU-Z supports windows 98.
Much more interesting would be to see the lowest x64 score.

I tried CPUz 1.73 on win98 and it didn't work, some Kernel error as expected, and I tried it without SSE on Win XP and it worked

but it's a little unreliable to say the least, 0 MT score, and ST score higher than a P3 800, even if their reference result was using the worst possible chipset it shouldn't be lower than a PII.

 

[dark zero]

Waiting for a U tier chip defeat this.
Well... I have enough laptop for 10 years at least.

 

[ArisVer]

Pentium B980 and Xeon E5450

 

[YBS1]

I found another little oddity with this benchmark while playing with my 3930K, or perhaps it's simply something with my settings, though I've not seen this happen before. I run my daily settings at 4.6GHz on it, I thought it odd my multithread was considerably higher than the reference 5930K yet my single thread was noticeably lower. The CPU idles at about 1200MHz under no load usually. I noticed in the single thread test it only ramps up to 4.2GHz, even after multiple runs or quick start/stop/start attempts to get it to jump. As soon as the multithread tests kick in all cores instantly fire to 4.6GHz. Odd, it's almost as if the system doesn't deem it enough load to ramp up fully for the single thread.

 

[rancherlee]

I ran my 8320@ 4.5 ghz several ways setting core affinity in Win10, using 1866 ram @ 9-10-9

4m/8c 1070st 6652mt
4m/4c 1059st 3115mt
2m/4c 1061st 2884mt
2m/2c 1061st 1028mt
1m/2c 1069st 928mt
1m/1c 1067st 3mt (yes, 3)

It appears to only give a slight penalty on the AMD FX for using 2 full modules vs. 4 modules with a core disabled on each one.

 

[LTC8K6]

1715 / 6261

4790K 4.4 2133 ram

1752 / 6408 at 4.5Ghz

2133 CL10 ram

 

[DrMrLordX]

might be intentional to make a well rounded bench. a cpu's ability to juggle two threads at once on the same core is also indicative of performance

Doubtful. There's a post earlier in this thread mentioning that it is a bug in the program's thread pool.

You also were on to something historically.
Does not having SSE instructions, mean the CPU can't run this bench at all or will it perhaps run it slowly with MMX or whatever?

It just means it'll fall back onto x87 fpu.