WINRAR: Benchmarking
Thread poster: DZiW

DZiW
Ukraine
English to Russian
+ ...
Feb 28, 2011

Recently I read some posts stating that there's not really much difference between many PCs built since some 2005 and modern ones (but videocard). For example, they say that WINRAR benchmarking (shortcut ALT+B) tests CPU (real and virtual cores), memory and HDD gives about 1000 points for many PCs made in 2005 and about the same for PCs made recently (except mobile solutions which traditionally fall behind). It assumes or at least implies that p4 3.2 x32 / DDR-400/ IDE (978 points) is roughly equal to modern systems with Duo/Quadro+ x64/ DDR2-1066/ SATA-2 300 (1022 points)...

Well, I realize that multi-core is about improved stability and responsibility only, that 64-bit is not 2x 32-bit, that slower RAM timings may work even faster (esp. when buffered and dual-channeled), that HDD speed is up to 60GB/s despite 133/150/300GB throughput which is mostly for RAID and yes--I'm aware that any system works as fast as the slowest system device or component; but why so little difference? I checked a dozen of different PCs and this statement seems plausible.

So, although the test is purely synthetic I would like to learn more about the fact of the matter: whether PCs are not really modernized much -OR- it's some software/ algorithm issue?

TY


Direct link Reply with quote
 

FarkasAndras
Local time: 03:35
English to Hungarian
+ ...
How so? Feb 28, 2011

DZiW wrote:

Recently I read some posts stating that there's not really much difference between many PCs built since some 2005 and modern ones (but videocard). For example, they say that WINRAR benchmarking (shortcut ALT+B) tests CPU (real and virtual cores), memory and HDD gives about 1000 points for many PCs made in 2005 and about the same for PCs made recently (except mobile solutions which traditionally fall behind). It assumes or at least implies that p4 3.2 x32 / DDR-400/ IDE (978 points) is roughly equal to modern systems with Duo/Quadro+ x64/ DDR2-1066/ SATA-2 300 (1022 points)...

That doesn't make much sense to me. Forget 5 years ago, current SSDs absolutely trounce every hard drive that was available even 1 year ago. Corei7 processors walk over the old core 2 duos, let alone whatever intel was shipping in 2005-2006. Moore's law still applies.

DZiW wrote:
Well, I realize that multi-core is about improved stability and responsibility only

I don't know where you got that from. Multi-core architectures are for multithreaded uses. If you're multitasking (watching a flash video and doing something else at the same time etc), they provide a huge performance boost. They can even double (quadruple) performance for single applications if those applications were coded well.

DZiW wrote:
that HDD speed is up to 60GB/s despite 133/150/300GB throughput which is mostly for RAID

I'm sure you mean MB/s. Good consumer SSDs are well above 200MB/s, with the new generation announced a couple of days ago hitting the 500 mark in both reads and writes. Due to the SSD explosion, storage is the fastest developing area in consumer computer tech now. I wouldn't be surprised if these monsters gave 2005-vintage RAM a run for its money in sequential reads.


To sum up: if this benchmark shows no significant improvement for the last 5 years, the benchmark is wrong. It's probably a primitive affair that was never meant to be used as a basis for the comparison of general performance; winrar is not benchmarking software after all.
Of course, whether the average user really needs the extra firepower - and what software developers are doing with it - is another story.

This post puts your post made ages ago about switching off Windows updates into perspective...

[Edited at 2011-02-28 23:11 GMT]


Direct link Reply with quote
 

DZiW
Ukraine
English to Russian
+ ...
TOPIC STARTER
not so fast Mar 1, 2011

Thank you, FarkasAndras.

Actually I wanted a qualified or at least sensible reply, but your promo will also do
First of all, I didn't mention any SSD, SAS-SCSI, DDR3-1666, RAM disks, CPU prototypes or real multiprocessor architecture and so on. Mind it.

It was about:
a ) old uni-, duo and 'HT'(hyperthreading/transport) CPUs -vs- modern multicores;
b ) old x32 -vs- modern x64
c ) old DDR400 -vs- DDR2-667-800-1066;
d ) IDE -vs- SATA -vs- SATA-2;
And modern RAR benchmarktest (with multicpu/core support) which uses the above.

As for multi-core CPUs (a low-budget version of a real multiprocessor approach) then I can surely know that 2-core CPUs is not as fast as 2 separate CPUs, is it clear? Actually, the performance gain (for specific multi-core aware software) is up to 30-50% (yet often may even slow down badly) and mostly for background processing, as you did mention. Thus, multicore CPUs have only their 'responsiveness' improved, not the overall speed-up. To tell the truth, it's a very common delusion very related to the nonsense made-up 'facts' like a 2000MHz CPU is as fast as two 1000MHz CPUs, that 64-bit machine is working at speed of two 32-bit ones, or that SATA2-300 is as fast as two SATA-150 (what could be valid for RAIDs only sharing the same controller). The real difference is the videocard...

Yep, of course, I meant MB/s, my bad) It's a pity that your post is rather beside the point, yet I see no use in arguing for you are but jumping to your 'conclusion' via irrelevant comparison: I asked what I asked and if one knows or has a relevant idea then I'm glad to hear it. Indeed, it has little to do with how you update your system or whatever.

So, is the machine that cost one arm and leg is really much faster or vendors just charge a top dollar or two for nothing?
- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -

Ok, here's a prompt reply from the RARLAB:
WinRAR "Benchmark" does not utilize HDD at all. It performs in memory
compression of randomly generated data. So it is not completely
synthetic, because it performs the real data compression.
But also it is not completely natural, because those randomly
generated data may be not that similar to typical data sets processed
by typical end user.

The real bottleneck for RAR compression is the speed of random access
to memory. Not established read/write speed in large blocks,
but random access to small data objects, probably not even cached.
This parameter is not improved as greatly as other parameters of
modern computer.Thanks guys. It's setlled)


[Edited at 2011-03-01 13:51 GMT]


Direct link Reply with quote
 

Jerzy Czopik  Identity Verified
Germany
Local time: 03:35
Member (2003)
Polish to German
+ ...
Do I need a benchmark to notice my machine is to slow? Mar 1, 2011

I don't think so.
Comparing the machines we have here you can clearly differ, which one is slow and which one does comply to my nowadays needs.
While obviously I started with a 12 MHz X286 processor, 640 kB RAM and a "huge" 40 MB HDD and it was fast enough I now notice, that a dual core Pentium 2 GHz laptop with 3 GIG RAM, 250 GB HDD and some Intel graphics is far too slow for my needs.
Needles to say, that a netbook with Intel Atom 1,33 GHz is not able to provide much more than a nice experience of having it with you wherever you want and have acces to emails (it is just 640 g in weight).
But I do not have any benchmarks for the machines nor do I really care about those.
My experience is, that after some 3-5 years I need a new PC - the old one does not fit anymore.


Direct link Reply with quote
 

FarkasAndras
Local time: 03:35
English to Hungarian
+ ...
laying this to rest Mar 1, 2011

DZiW wrote:

Actually I wanted a qualified or at least sensible reply, but your promo will also do

Let me know what I'm promoting so I can go pick up my check from whoever it is.

DZiW wrote:
First of all, I didn't mention any SSD, SAS-SCSI, DDR3-1666, RAM disks, CPU prototypes or real multiprocessor architecture and so on.

I didn't mention any exotic components, either.
SSDs are not the stuff of fairytales anymore, they are standard in high-performance setups, even high-end consumer-oriented laptops and so on. You can get the sort of performance I was talking about for $110.
Multicore processors, which is what I was talking about, are standard in everything but the lowest of the low end. I'm pretty sure dual-core processors have separate silicon for each core and work pretty almost exactly like two processors, so there's no need to make a distinction between multicore and "real multiprocessor" setups.


DZiW wrote:As for multi-core CPUs (a low-budget version of a real multiprocessor approach) then I can surely know that 2-core CPUs is not as fast as 2 separate CPUs, is it clear? Actually, the performance gain (for specific multi-core aware software) is up to 30-50% (yet often may even slow down badly) and mostly for background processing, as you did mention. Thus, multicore CPUs have only their 'responsiveness' improved, not the overall speed-up.

I'd say a 30-50% performance bump is rather significant.
I'd also say that improved responsiveness is improved performance. If my computer does what I want it to do instantly instead of 2 seconds later (when it's done with some other task, which might be some OS-related system task that runs in the background), it's working faster. Operations you want to do get done faster, I don't see how that's not a performance improvement.

As to the rest, you got your answer both from me and winrar: this benchmark wasn't intended to provide a realistic and comprehensive comparison of various hardware setups.


Direct link Reply with quote
 

Phong Le  Identity Verified
Vietnam
Local time: 08:35
Vietnamese
+ ...
Faster or not, should be based upon the same tasks Mar 2, 2011

All,

I understand the original question. Seems 05 years ago, you wait for 2 min to start up a PC. Now it is the same time. So no improvement?

If we use the new PC i3 core, and set up windows 98 with office 97, am sure that it is much faster than the P3 core.

One of the issue is the operation system and the software are updated and asked for more resources to run it.

Then the hardware is updated to a certain level and then the software again updated higher...

Phong


Direct link Reply with quote
 

DZiW
Ukraine
English to Russian
+ ...
TOPIC STARTER
it's clear Mar 2, 2011

that some PC components are really improved much, yet some are not that much.

I didn't say a word in my original post about SSD because I've got a couple in my net- and notebooks (yet prefer SaS for desktops), so I'm aware what they are. But I was mistaken when thought (as some posters) that HDD was considered in WINRAR Benchmarking & Hardware Test too. Thus, as far processing datastreams with random access is almost equal to those PCs 5 years ago I thought there was something wrong and I war right this time

FarkasAndras, considering multicore and multicpu as the same is like confusing translators and interpreters, or thinking school and uni level of education is 'almost' equal, ain't it an education after all? In fact it's but economical gimmick for real CPUs are real fast, not fake 'cores'. Sure, multicore speeds up the processing, but mostly for background, not foreground software which also must be multi-core aware. No, I guess you're mistaken thinking that system 'responsiveness' (smoothness) is the real performance improvement for it doesn't speedup the foreground task much.

Be more attentive: It performs in memory compression of randomly generated data. So it is not completely synthetic, because it performs the real data compression. So it is the real test, but I'd say it's narrow or specific-purpose, ain't you compress anything even on your CoreI7 or Opteron?

Yep, once I found an old notebook with p3/512 and I was surprized that it worked perfectly well and running MS Office XP under XP SP3 x32 was quite smooth.
Ok, now I know the correct answer and don't care much who uses what and how

Cheers.


Direct link Reply with quote
 

FarkasAndras
Local time: 03:35
English to Hungarian
+ ...
Here's an idea Mar 2, 2011

Show me a "multiprocessor" consumer PC and I'll take your proclamations about multiprocessor vs multicore setups seriously. I don't see why you think two processors would be significantly better than two cores in the same chip, honestly. I'm sure otimizing load distribution would be a huge pain... but it's a moot point because it's an imaginary setup anyway. Why you think the individual cores in multicore CPUs are "fake" is a mystery to me. Perhaps you are confusing multi-core processors with hyperthreading and virtual cores, I don't know. I don't feel like untangling your posts any more, to be honest.

If you think multi-core processors offer no tangible improvement over single-core architectures, well, you have a right to that opinion. You won't find many people sharing that opinion with you, but don't let that bother you.


Direct link Reply with quote
 


To report site rules violations or get help, contact a site moderator:


You can also contact site staff by submitting a support request »

WINRAR: Benchmarking

Advanced search






Protemos translation business management system
Create your account in minutes, and start working! 3-month trial for agencies, and free for freelancers!

The system lets you keep client/vendor database, with contacts and rates, manage projects and assign jobs to vendors, issue invoices, track payments, store and manage project files, generate business reports on turnover profit per client/manager etc.

More info »
WordFinder Unlimited
For clarity and excellence

WordFinder is the leading dictionary service that gives you the words you want anywhere, anytime. Access 260+ dictionaries from the world's leading dictionary publishers in virtually any device. Find the right word anywhere, anytime - online or offline.

More info »



Forums
  • All of ProZ.com
  • Term search
  • Jobs
  • Forums
  • Multiple search