Hello Beemerke, welcome to Seven Forums!
Good score! If you don't have AHCI already enabled in the BIOS (and you have that option) you may get a better number with that Intel SSD with AHCI enabled; have a look at the tutorial at the link below; be sure to post back to keep us informed.
AHCI : Enable in Windows 7 / Vista
Thanx for the welcome :)
What setting needs to be changed in the bios, is it the RAID one (from ide to ahci)?
I played with the settings a bit, windows did install drivers but the WEI score stayed the same @7.2, should be a bit more for an X25-M 80GB SSD, guess Snakeyeskm was right
I have another problem I hope you can help me with, not sure where to post it so I'll post it here, delete this and post it in the right section when I get an answer where to post it.
I made a new toolbar to put all my icons in and also to have a functional recycle bin on my taskbar (cleanup the desktop). When I close my computer and boot the next day the toolbar is gone, this does not happen when I reboot.
When I make a new toolbar all the buttons I put inthere the previous day are in the new toolbar so I guess is stored somehow but the toolbar itself doesn't show up on the taskbar automaticaly. Anyone have an idea what's going on?
Ok. I ran the Futuremark Vantage benchmark. The numerical score is 7707...whatever that means.
The quick summary image is attached.
When I requested a comparison, the Futuremark site informed me that there was only one other system even remotely like mine and no valid comparison could be made.
mbreslin: I understand your comment about the disk subsystem being the usual suspect for a bottleneck. I addressed that in a previous post, but just to give a quick overview, I can't take the time nor expense at the moment to install a huge RAID 3 or 4 array at this time. It's the beginning of my busy season and that won't end until late October or early November. Maybe by then I'll be frustrated enough to go buy 4 of the 2 tb drives and use Raid striping to emulate multiple paths to disk.
I think one of the reasons that HDDs are a bottleneck is the use of SATA interfaces. I understand that the data transfer rate is greater on SATA 2 than IDE, but what people seem to overlook is that while IDE may only run at a clock rate of 133 mb/sec, there are 16 lines transferring data in parallel, whereas with a SATA 2 interface there's 1 (one) line transferring 300 mb/sec.
This serial/parallel issue goes back to the 70s when some disk pack drives used parallel interfaces (and huge cables) and other drives used serial interfaces...their cables weren't as big. But when put into real world situations, it was found that the allegedly higher clock rate serial transfer drives required 2 channels from the controller, 1 for read, 2 for write to equal the performance of the parallel drives. Then it got ugly when external controllers and I/O exchanges were added to do load shifting to the least used path. It made troubleshooting a nightmare. I came to hate the serial interface drives.
As for Intel vs AMD, to use an analogy in another field I'm familiar with, cameras, there are 2 front runners for digital single-lens reflex cameras: Nikon and Canon. I use Canons. Until a short time ago Canon had the best DSLRs on the market. Now Nikon has introduced the D3 and will be releasing the D4 which will probably put them at the top of the heap. I don't see any impending releases from Canon, but if things follow their usual course of events, they'll release a 1D mk something or a 5D mk 4 and the war will continue to escalate.
I perceive the same thing happening with AMD and Intel. Yeah, the Intel CPUs are the hot ticket right now. But my guess is that the pendulum will swing the other way soon.
However, back in my initial post I wrote that I wasn't trying to build the fastest, newest machine possible. I wanted (and still do) a machine that's more like a 5-ton truck rather than a Porsche. I wanted to be able to run a lot of applications without severe performance degradation. That was the reason for using 2 quad-core chips. If I'd wanted something blindingly fast in a single-threaded setup, I could have taken a different route and cooled it with dry ice just to get the highest overclock rate. But that wouldn't get my photos edited any faster, nor would it help with web site development. In my case, it's not purely speed I'm looking for, it's stamina. I believe I built a box that should have produced that result, but hasn't. That's why I'm here looking for something that I may have overlooked or a suggestion to help me reach that goal.
@raytherat: I am very sorry I completely forgot about that benchmark not really taking advantage of SMP systems. That score is horribly low but I'm sure a lot of it has to do with your having a multi-proc system. If I had to guess I would say most of the software you use doesn't fully take advantage of your two processors, I'd bet this causes the majority of your woes.
I'm afraid I'm out of ideas, your system is built I don't know quite what else to say than maybe try and tend towards software that is optimized for smp systems.
Storage interface: Command Queuing and the increase of bandwidth makes sata the clear winner over ide on platter drives. And with ssds access times are so low it is difficult to even reach high queue depth except in synthetic benchmarks so serial/parallel is moot.
Intel vs Amd: I see there being two pendulums, price versus performance, and raw speed. The pvp pendulum sometimes swings to intel but certainly spends the majority of the time for many years now in the amd camp. But as for raw speed intel nearly always has the fastest chip available. I can think of only twice in 10 years when amd had a chip that could beat intels latest in raw performance and the lead wasn't held very long.
I would like to note I hold no allegiances to any of these companies, as they certainly don't care about me, all of them will simply charge the very most they can get me to pay. I recently built this machine and I'd bet it does everything you'd want and more (no dry ice required), though I paid a nearly insane premium for it.
These are all just my opinions of course. In the case of your machine I honestly feel the problem is although SMP has been around for quite some time as it's the minority hardly any software is truly optimized for it. It's been around long enough that there are smp enabled versions of lots of apps but the difference between smp enabled and smp optimized is great.
mbreslin: I had to laugh at something. Not you, but as I was investigating how SMP architecture is implemented today, I found something in Wikipedia that's just too ironic not to share.
Symmetric multiprocessing - Wikipedia, the free encyclopedia
"The Burroughs B5500 first implemented SMP in 1961.[1] [2] It was implemented later on other mainframes. Mid-level servers, using between four and eight processors, can be found using the Intel Xeon MP, AMD Opteron 800 and 8000 series and the above-mentioned UltraSPARC, SPARC64, MIPS, Itanium, PA-RISC, Alpha and POWER processors. High-end systems, with sixteen or more processors, are also available with all of the above processors."
All of the mainframe experience I've been referring to has been about Burroughs systems, for whom I worked for 16 years, until they merged with Sperry and cut their world-wide workforce from 130,000 employees to 30k. That's when I went from hardware (and system software) into application development.
It's wonna them real small world things.
Here's what I've done. I've removed the swapfile. That freed up 16 gig on the 60 gig SSD. That's pretty significant. I tested the software that I generally use and it all works fine that way. Then I set the graphics mode to "performance" It looks clunky as hell, but I'm gonna try it this way. And I also switched the processor scheduling setting from Programs to Background services. That's the way it's running now.
Nothing's crashed and it seems to be running acceptably well.
But. (Why is there always a but?) I've kept ResMon open as I've been doing all this stuff (it's displayed on the secondary monitor) and I've never seen more than 3 gig of ram in use. Right now I have 13 applications, 929 threads and 21,500 handles assigned. Still, only 2 gig of ram is in use.
Oh. I happened to read on another forum that many people were finding Explorer to be a major bottleneck and some suggested it might be the source of the memory leak problem that keeps coming up. So I installed Free Commander. I have Xplorer2, but I don't like the fact that it can only have 1 tree view open. Free Commander allows tabbed exploring, side-by-side "panels", each with its own tree. Initial tests seem to be acceptable. I'm gonna have to go shoot a follow-up on the restoration of a Land Speed Racing Streamliner
Ray the Rat's Chevy Asylum so I'll be away for a while, but in the meantime, I'll let this thing percolate and see what it does. Matter of fact, I'm gonna tell it to do a huge thumbnail rebuild in ACDSee and see if it can actually complete it for once. I just fired it off and while it increased the processor load, memory use is still at 2 gig.
As far as your view of two pendula and their swings re: price and performance, I believe you have a very good point. They're often two separate and distinct issues. And I'll admit that I couldn't justify buying the very top of the line CPUs, nor could I go for something like a 1tb SSD. I shudder to think about those prices.
Back on SMP, you're right, although multi-core CPUs may invite more applications to optimize for that. However, there are a coupla things going on here. One is that the architecture of this system is 2 processors, with 4 cores each and two sets of 8 gig of RAM. So some of the commonly-held concepts about SMP working on a common memory stash aren't exactly true here. And there's one other issue to be considered. With a true multi-processor architecture, the OS SHOULD be able to perform load-leveling and assign single-threaded processes to a single core. I believe this is happening, since I can see activity reported on all 8 cores. So while parallel processing of multiple instructions with an SMP-optimized OS and application would be the best solution, load-leveling is probably next best.
And just to bash on my favorite issue, in all the years I've been working with computers that use volatile memory, almost all CPU performance issues have been helped by increasing the amount of available RAM. That's why I left XP. I had a box with 4 gig and XP could only see 2.
Anecdotal evidence seems to indicate that when someone complains that their computer is running slowly, one or more of 4 things is usually going on:
1: Too many applications loading at startup.
2: Insufficient RAM resulting in constant swapfile use. I wish I could remember the term we used for this, but in first generation mini-computers a system could go into gridlock when it kept swapping one segment out for another, then realizing it needed the one it had just swapped out the one it needed to complete the task on the presently loaded segment. I just remembered it. It was called (colorfully enough) "deadly embrace." and would put the system into an infinite loop of swapping memory segments in and out.
3: Malware masquerading as anti-spyware (this is only recent...that is the last few years)
4: Disk fragmentation.
There are several others, but these seem to be the most prevalent that I've seen.
In the first 2 scenarios, increasing RAM will generally be helpful. Certainly not in #3, but it could prevent #4 from occurring as frequently if swapping is minimized.
But I've never seen a dela like what I'm looking at now. I've attached a screen capture showing memory usage. Just weird.
Oh..the thumbnail rebuild that would take forever? It finished in less than 5 minutes. Maybe the setting for task scheduling has helped that.
I'm outta here to go shoot some photos.
@raytherat: Great points and I'm sorry I couldn't be more helpful. Best of luck.
Here's my score on a new i7 build. Grrrrrr....those darn mechanical HD's
