Introducing: MaxxPI² - PI based CPU/Memory benchmark

**DdotRoq** · 05-29-2009, 09:39 AM

Neat little program, a nice addition to this would be a simple core temperature tracker.

**alicemb** · 06-02-2009, 09:24 AM

hi,

now new as a part of MaxxPI²'s memory benchmark,
as preview-version (Memory/Latency, v1.05):

www.maxxpi.net

cu

**alicemb** · 06-18-2009, 11:22 PM

hi,

update MaxxPI² - PreView - Single to v1.41

• v1.41, Batchmode added (18/06/2009) NEW!

cu

**alicemb** · 07-15-2009, 10:53 PM

hi,

MaxxPI² - PreView - Multi

This benchmark uses as formula the Chudnovsky algorithm, unlike the MaxxPI ² - PreView - Single that use the Gauss-Legendre algorithm. The advantage of the Chudnovsky algorithm is, that principally, multi-core capability is possible. MaxxPI ² - PreView - Multi use this. That
means: That all available CPU cores work together on a single calculation. It was made sure that all CPU's equally served.

Technical:
MaxxPI² - PreView - Multi needs at least a dual-core processor and supports in the current version 1.07 CPU's with 2,3,4 and 8 Core's.
So the benchmark will only run on CPU's, with more than one core (HT core counts as real core, so 1+1HT core will accepted).

Maximum depth of calculation:
268.435.456 decimal places

•v1.07, initial public release (16/07/2009) NEW!

cu

**alicemb** · 07-24-2009, 07:08 AM

hi,

update MaxxMEM² - PreView to v1.07

* v1.07, improved CPU detection (24/07/2009) NEW!

download: http://www.maxxpi.net/pages/downloads.php

and MultiPI`s scaleability http://www.maxxpi.net/pages/descript...aleability.php

cu

**alicemb** · 09-10-2009, 06:14 AM

hi,

MaxxMIPS² - PreView v1.45
Pure CPU benchmark (CPU String/Integer -part, ALU)

v1.45, initial public release (10/09/2009) NEW!

cu

**KB24LA** · 09-10-2009, 06:27 AM

for how long internal beta testing will continue?
nice utilities btw

**alicemb** · 10-23-2009, 01:06 AM

hi,

update MaxxPI² - PreView - Single :

# v1.60, major algorithm change from Gauss–Legendre to Chudnovsky (16/10/2009) NEW!

for much better scaling on *all* cpu types.

www.maxxpi.net

cu

**alicemb** · 12-02-2009, 11:12 AM

hi,

for the intressed reader:

Memory all inclusive view

Mips scaleability

Flops scaleability

cu

**Duh** · 12-02-2009, 01:51 PM

Originally Posted by mattkosem

I see 481.5 K/sec for 1M on my e8600 @ 4.53Ghz (533x8.5).

--Matt

phase change?

**FireDragon** · 12-13-2009, 10:08 PM

these seem like nice programs and they stress the cpu a fare bit unlike super pi, they could b a nice stability tester i will look more into them for that

Dragon

**alicemb** · 12-21-2009, 08:27 AM

hi,

comming soon: first HWbot step, of 2(3)

best

**alicemb** · 12-30-2009, 12:06 PM

update,

MaxxMEM²-PreView:

* v1.73, MaxxPI.net / HWbot.org submit added (30/12/2009)

**alicemb** · 01-03-2010, 02:37 AM

update for:

MaxxPI²-Multi-PreView
* v1.55, support for upcomming new 6/12 core CPU's added (03/01/2010) NEW!

MaxxMEM² - PreView
* v1.74, minor bugfixes (03/01/2010) NEW!

Download

cu

**alicemb** · 01-23-2010, 04:34 AM

hi,
for interested readers:

Vcore characteristics

cu

**Necetra** · 01-23-2010, 05:40 AM

Originally Posted by alicemb

hi,
for interested readers:

Vcore characteristics

cu

That's a very, very cool chart. Gonna have to try this program out when I can get my ass back to my desktop.

**mattkosem** · 01-23-2010, 07:56 PM

Originally Posted by Duh

phase change?

Na, that was just water. Here's what I get on water with my i7 at 24/7 clocks.

--Matt

**mattkosem** · 02-06-2010, 10:58 PM

This benchmark doesn't seem to benefit much from HT.

--Matt

**alicemb** · 02-07-2010, 02:28 AM

Originally Posted by mattkosem

This benchmark doesn't seem to benefit much from HT

sure it does. HT benefit is only then possible, if the used routine don't use *much real* CPU power, this is *not* seen with classic taskmanager.
Performance Monitoring Unit (PMU) is the key. if there is a very high useage
(maxxpi peaks there about >85%) then there is nothing for HT to give for.

and dont think that HT will give you twiche performance

normally enabling HT at i7 9xx will give you a plus ~400-600mhz and that quite fine for *virtual* cores

best

**alicemb** · 02-07-2010, 02:38 AM

Originally Posted by Necetra

That's a very, very cool chart. Gonna have to try this program out when I can get my ass back to my desktop.

yes it is. thank you, but this three:

Memory all inclusive view

Mips scaleability

Flops scaleability

are also very nice

best

**mattkosem** · 02-07-2010, 09:59 AM

Originally Posted by alicemb

sure it does. HT benefit is only then possible, if the used routine don't use *much real* CPU power, this is *not* seen with classic taskmanager.
Performance Monitoring Unit (PMU) is the key. if there is a very high useage
(maxxpi peaks there about >85%) then there is nothing for HT to give for.

and dont think that HT will give you twiche performance

normally enabling HT at i7 9xx will give you a plus ~400-600mhz and that quite fine for *virtual* cores

best

I think you misunderstood what I was saying. It definitely takes advantage of it, the scores just don't benefit much from it. 4.5Ghz without HT is faster than 4.2Ghz with HT, even with the 4.5Ghz test running on lower memory and uncore clocks.

--Matt

**alicemb** · 02-07-2010, 11:09 PM

Originally Posted by mattkosem

I think you misunderstood what I was saying.

if its so, sorry for that. english is not my native

Originally Posted by mattkosem

It definitely takes advantage of it, the scores just don't benefit much from it. 4.5Ghz without HT is faster than 4.2Ghz with HT, even with the 4.5Ghz test running on lower memory and uncore clocks.

well, not *much* right,
because as i said: Monitoring Unit (PMU) is the key.
if there is much load on this, HT will give you not much (because there is nothing to give for).
in worst Case: *lower* performance then without.

best

**alicemb** · 04-19-2010, 12:27 AM

hi,

update for:

MaxxMEM² - PreView

* v1.80, maintenance update, Quick compare added (19/04/2010)

Remark for Quick compare:
with this, you can very fast see how your memory subsystem
perform-> red-yellow-green.
based on the MaxxPI DB, which contains low-end PC's
and highly/extreme overclocked ones.

cu

cu