 |
 | Home | Contact Us | FAQ | Search & Site Map | Link to Us |
 | Sign In | Join | Other 45 Sites in Network |
| HomeDiscussionsAccessExcelInfoPathOutlookPowerPointPublisherWordDirectoryUser Groups Related Topics Outlook ExpressInternet ExplorerWindowsMS Server ProductsMore Topics ... |
MS Office Forum / Excel / New Users / October 2006Hardware specification for dealing with large files (Excel 2003 and 2007)
Hi All Apologies for repeating parts of other subjects, but I was wondering what people thought were the most important hardware requirements for a machine to effectively calculate large Excel spreadsheets. Specifically, I have problems at the moment both with re-calculations (can take a few minutes) plus things like re-calculating cell contents (when copy-and-pasting formulae) can take forever, often leading me to kill the Excel process.. We're currently on Excel 2003 but are looking to upgrade when the new version comes out. As a starter for ten, I'm guessing the following are the important factors: 1) Processor - as fast as I can get. Is it worth getting a dual processor? I read somewhere, "No" for Excel 2003 but possible for 2007. 2) RAM - as much as possible. I'm thinking 2Gb. 3) Hard disk - not as important because I should, hopefully, have enough RAM to deal with the spreadsheet. 4) Display cards etc - more or less irrelevant. Thanks Ben Ben, the processor is important. Do NOT get a Celeron, because they do not have math co-processors. Because of this, a real Pentium rated slower than a Celeron will be faster than the Celeron in crunching numbers. James > Hi All > [quoted text clipped - 20 lines] > > Ben James; I'm just curious; do you have some evidence for that? I think that it sounds like a load; personally. I'd just love to find out more about it. For example; would a math co-processor help a database server? Back in the day; these were on the motherboard; and not the processor.. right? Any way to make a db server faster; by adding a math co-processor card? just curious -Aaron > Ben, the processor is important. Do NOT get a Celeron, because they do > not have math co-processors. Because of this, a real Pentium rated [quoted text clipped - 25 lines] > > > > Ben jkend69315@aol.com wrote... >Ben, the processor is important. Do NOT get a Celeron, because they do >not have math co-processors. Because of this, a real Pentium rated >slower than a Celeron will be faster than the Celeron in crunching >numbers. ... Celerons don't have co-processors per se. They have built-in floating point units (FPUs), which have been part of Intel CPUs since the 80486DX. The FPUs process the machine instructions formerly processed by math co-processors (through the 80387). I got talked into a Celeron against my better judgement when I first got this very machine built. It just crept along on doing number crunching. It was so slow I had to take it back and get a real Pentium installed. The difference was amazing. I was told the difference was due to the Celeron's lack of a math coprocessor, so that may not be exactly right. But I can say from experience a Celeron is not a good choice for number crunching. James > jkend69315@aol.com wrote... > >Ben, the processor is important. Do NOT get a Celeron, because they do [quoted text clipped - 7 lines] > 80486DX. The FPUs process the machine instructions formerly processed > by math co-processors (through the 80387). it's just fascinating. i've never had a celeron. is the same thing true of 'semprons and durons' vs athlons? > I got talked into a Celeron against my better judgement when I first > got this very machine built. It just crept along on doing number [quoted text clipped - 15 lines] > > 80486DX. The FPUs process the machine instructions formerly processed > > by math co-processors (through the 80387). Based on the above I'll avoid a Celeron but - what about the issue of a dual processor? As I say, I read somewhere that Excel 2003 will only use one processor, making it pointless to have two, but that Excel 2007 will provide this functionality. Also though, surely it would help having two, even with Excel 2003, because the other processor could, theoretically, deal with all the other stuff (Email, Word, Internet, all the other processes that seems to run on a modern PC which I have no idea what they does), whilst the first processor looks after number crunching.. Ben Ben, My dismal experience with the Celeron is about the limit of my experience on hardware issues. Seems like two processors would be better than one -- the two heads better than one axiom -- but I have to admit ignorance. I guess everyone knows that you just can't have too much RAM. And there are graphics accelerators that are good for that kind of thing. And of course, laptop processors are a different animal. Aaron, I haven't had an AMD processor since the old K6 days. The K6 was a very good product. But the shop I've been using for my PCs for the last few years uses Intel. I wish I knew more about hardware issues. There must be a newsgroup out there somewhere on this subject. James > Based on the above I'll avoid a Celeron but - what about the issue of a > dual processor? As I say, I read somewhere that Excel 2003 will only [quoted text clipped - 7 lines] > > Ben benjrees@gmail.com wrote... >Based on the above I'll avoid a Celeron but - what about the issue of a >dual processor? As I say, I read somewhere that Excel 2003 will only [quoted text clipped - 5 lines] >no idea what they does), whilst the first processor looks after number >crunching.. It's true that Excel 2003 and prior won't use multiple processors. However, that implies that multiple processor support is built-in at application rather than OS level, and that would mean that Windows can't use one processor for Excel and another for everything else. The OS, Windows, may use both processors, but I believe it only provides one set of time slices. That is, Windows won't run one application on one processor and others on the other processor at the same time. Rather, Windows runs all applications in repeating sequence, giving a few milliseconds of processing time to App1, then a few to App2, then a few to App3, then back to App1, etc., and when any of these apps is running, it can make use of multiple processors or not. IOW, Excel 2003 and prior derive NO BENEFIT from running on systems with multiple processors. Other applications may, but not from running in parallel on one processor while Excel runs simultaneously on the other processor. Excel 2007 supposedly can spread processing over multiple processors. jkend69315@aol.com wrote... >I got talked into a Celeron against my better judgement when I first >got this very machine built. It just crept along on doing number [quoted text clipped - 3 lines] >exactly right. But I can say from experience a Celeron is not a good >choice for number crunching. James ... The big difference between Pentium and Celeron is the on-chip cache: the Pentium's is larger than the Celeron's. For some types of number crunching, that could be critical. Cache faults require memory access, and that in turn involves differences in bus speeds between Pentiums and Celerons. again; i'm just curious.. do Durons and Semprons have the same problem; are they missing a math co-processor? it could just be that I've had performance problems doing trivial tasks in Excel because I'm not aware of these hardware requirements for excel. -Aaron > jkend69315@aol.com wrote... > >I got talked into a Celeron against my better judgement when I first [quoted text clipped - 11 lines] > and that in turn involves differences in bus speeds between Pentiums > and Celerons. aaron.kempf@gmail.com wrote... >again; i'm just curious.. do Durons and Semprons have the same problem; >are they missing a math co-processor? ... Check the processor specs online. Google search on Duron or Sempron. With regard to Celerons, they have on-chip FPUs. That's *NOT* how they differ from Pentiums. They differ in terms of on-chip cache size and clock speed/bus speed differentials. thanks Harlan.. glad to see correction of mis-information I was curious; because I haven't heard of that math co-processor explanation before so it's not like it's a 10:1 performance difference; it's like a 20% performance difference right? > aaron.kempf@gmail.com wrote... > >again; i'm just curious.. do Durons and Semprons have the same problem; [quoted text clipped - 6 lines] > differ from Pentiums. They differ in terms of on-chip cache size and > clock speed/bus speed differentials. Harlan, Interesting stuff. Thanks for sharing your knowledge. I interpret your reply to mean that if the processor cannot cache sufficient data to perform its current task(s), it must resort to the bus to communicate with RAM or other memory external to the chip. And, in this case, the processor's speed in handling bus fetches will affect how much this operation slows down the process. Regards, James > aaron.kempf@gmail.com wrote... > >again; i'm just curious.. do Durons and Semprons have the same problem; [quoted text clipped - 6 lines] > differ from Pentiums. They differ in terms of on-chip cache size and > clock speed/bus speed differentials. aaron.kempf@gmail.com wrote... >again; i'm just curious.. do Durons and Semprons have the same problem; >are they missing a math co-processor? ... Check the processor specs online. Google search on Duron or Sempron. With regard to Celerons, they have on-chip FPUs. That's *NOT* how they differ from Pentiums. They differ in terms of on-chip cache size and clock speed/bus speed differentials. I googled duron and sempron and i got a lot of hits I can get general performance benchmarks on www.anandtech.com www.tomshardware.com I just was hoping for some more specific hardware specs using Excel; I mean-- since MS is getting rid of the limit on Rows and Columns; I assume that people are really going to start bitching when Excel chokes on 70,000 records lol -Aaron > aaron.kempf@gmail.com wrote... > >again; i'm just curious.. do Durons and Semprons have the same problem; [quoted text clipped - 6 lines] > differ from Pentiums. They differ in terms of on-chip cache size and > clock speed/bus speed differentials. aaron.kempf@gmail.com wrote... ... >I just was hoping for some more specific hardware specs using Excel; I >mean-- since MS is getting rid of the limit on Rows and Columns; I >assume that people are really going to start bitching when Excel chokes >on 70,000 records lol It'll choke well before that. The on-chip cache is unlikely to hold more than a few thousand rows from spreadsheets OR records from databases. 16mb cache on the new xeons.. right? I swear; if it cost $4 grand?? probably not that bad of an investment. > aaron.kempf@gmail.com wrote... > ... [quoted text clipped - 6 lines] > more than a few thousand rows from spreadsheets OR records from > databases. |
Reply to this Message |
 Sign In
 Join
 My Latest Posts My Monitored Threads My Blog My Photo Gallery My Profile My Homepage Start New Thread Enable EMail Alerts Rate this Thread
|
©2008 Advenet LLC Privacy Policy - Terms of Use
This website includes both content owned or controlled by Advenet as well as content owned or controlled by third parties.