jmoore@cidmac.ecn.purdue.edu (James D Moore) (07/06/90)
I was asked to find out some information for a professor here at the university. A project we are doing requires nothing more than as much speed as possible from the processor. I realize that this is related to disk acess and such but the basic question I need an answer for is "Which is faster a 386/33mhz or a 486/25mhz?" Thanks!! Jim Moore -- +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+- James D. Moore jmoore@cidmac.ecn.purdue.edu Computer Engineer Phone:(317) 494-2686 Industrial Engr. Dept., Purdue University, W. Lafayette, In 47907
davidsen@sixhub.UUCP (Wm E. Davidsen Jr) (07/07/90)
Newsgroups: comp.sys.ibm.pc,comp.sys.ibm.pc.misc,comp.sys.ibm.pc.hardware Subject: Re: Whis is fastest 386/33 or 486/25 ? Summary: Expires: References: <1990Jul5.205440.23370@ecn.purdue.edu> Sender: Reply-To: davidsen@sixhub.UUCP (bill davidsen) Followup-To: Distribution: na Organization: *IX Public Access UNIX, Schenectady NY Keywords: In article <1990Jul5.205440.23370@ecn.purdue.edu> jmoore@cidmac.ecn.purdue.edu (James D Moore) writes: | is related to disk acess and such but the basic question I need an | answer for is "Which is faster a 386/33mhz or a 486/25mhz?" A 486 is about 2.6 times faster than a 386 at the same clock, for *most* integer calculations. It is more than that for floating point. Why stop at 25MHz on the 486, the 33MHz machines seem to be shipping. Even as stated, the 486 is noticably faster. -- bill davidsen - davidsen@sixhub.uucp (uunet!crdgw1!sixhub!davidsen) sysop *IX BBS and Public Access UNIX moderator of comp.binaries.ibm.pc and 80386 mailing list "Stupidity, like virtue, is its own reward" -me
kthompso@entec.Wichita.NCR.COM (Ken Thompson) (07/10/90)
Jim, The 486/33Mhz is faster. NCR is selling them. -- Ken Thompson N0ITL NCR Corp. 3718 N. Rock Road Wichita,Ks. 67226 (316)636-8783 Ken.Thompson@wichita.ncr.com
ggraef@csd4.csd.uwm.edu (gerald graef) (07/11/90)
In article <1990Jul5.205440.23370@ecn.purdue.edu> jmoore@cidmac.ecn.purdue.edu (James D Moore) writes: > >I was asked to find out some information for a professor here at >the university. A project we are doing requires nothing more than >as much speed as possible from the processor. I realize that this >is related to disk acess and such but the basic question I need an >answer for is "Which is faster a 386/33mhz or a 486/25mhz?" > >Thanks!! > >Jim Moore > There is always substantial deviation between manufacturers, but given equivalent motherboards (say, with external caches etc.) the fastest 80x86 is the 33mhz 486, followed by the 25mhz 486 and then the 33-386's. A 486 will run in the range of 2 times faster than an equivalent speed 386. -- --Common sense is the collection of prejudices aquired by age 18 - Albert E. --Only by purest chance do the above resemble the views of anyone other than: Gerald Graef: Internet %%%%% ggraef@csd4.csd.uwm.edu : BITNET %%%%% ggraef%csd4.csd.uwm.edu@INTERBIT
kdq@demott.COM (Kevin D. Quitt) (07/11/90)
In article <4952@uwm.edu> ggraef@csd4.csd.uwm.edu (gerald graef) writes: >In article <1990Jul5.205440.23370@ecn.purdue.edu> jmoore@cidmac.ecn.purdue.edu (James D Moore) writes: >> >>I was asked to find out some information for a professor here at >>the university. A project we are doing requires nothing more than >>as much speed as possible from the processor. I realize that this >>is related to disk acess and such but the basic question I need an >>answer for is "Which is faster a 386/33mhz or a 486/25mhz?" >> >>Thanks!! >> >>Jim Moore >> >There is always substantial deviation between manufacturers, but given >equivalent motherboards (say, with external caches etc.) the fastest >80x86 is the 33mhz 486, followed by the 25mhz 486 and then the 33-386's. >A 486 will run in the range of 2 times faster than an equivalent speed >386. I'm not sure how you derived this, unless you were dealing with clock speed only. The 486 is 2-3 times faster than the 386 on identical clock speeds (if Intel and a number of reviewers are to be believed). This being the case, the 486/25 must be significantly faster than the 386/33. And in fact, I have seen benchmark results that tend to indicate that it is faster. -- _ Kevin D. Quitt demott!kdq kdq@demott.com DeMott Electronics Co. 14707 Keswick St. Van Nuys, CA 91405-1266 VOICE (818) 988-4975 FAX (818) 997-1190 MODEM (818) 997-4496 PEP last 96.37% of all statistics are made up.
lowey@herald.usask.ca (Kevin Lowey) (07/11/90)
From article <1990Jul5.205440.23370@ecn.purdue.edu>, by jmoore@cidmac.ecn.purdue.edu (James D Moore): > > I was asked to find out some information for a professor here at > the university. A project we are doing requires nothing more than > as much speed as possible from the processor. I realize that this > is related to disk acess and such but the basic question I need an > answer for is "Which is faster a 386/33mhz or a 486/25mhz?" I have a benchmark program that I wrote which does a lot of tests including the sieve, fibonacci, whetstone, integer math, floating math, trancendental math, etc. It was done in Turbo Pascal 3.0 so the compiler does no math optimizations, etc. Here are some of the final results (in seconds) which I obtained. Machine Name CPU MHz NO 80x87 With 80x87 IBM Model 80/486 486 25 26.80 10.92 DEC 325c 386 25 46.96 Zenith 386 386 25 51.25 20.93 IBM P70 386 20 74.92 IBM PS/2 80 386 16 91.66 Compaq 386 386 16 93.07 57.35 Primax 316SX 386sx 16 96.73 IBM PS/2 model 60 286 10 163.89 60.59 IBM PS/2 model 60 286 10 170.03 65.07 (OS/2 1.1 DOS box) IBM-AT 286 6 281.65 AT&T 6300 8086 ? 365.64 IBM-XT 8088 4.77 783.56 281.11 Amiga 2000 8088 4.77 797.41 (bridge card) Mac IIx 68030 16 1309.91 (Soft-PC 1.21 DOS emulation) That should give you a few comparisons. Unfortunately I haven't tested any 33Mhz 386 boxes. But at the same clock speed, the 486 chip appears to be about twice as fast as the 386 chip (in real mode at least) - Kevin Lowey
laughner@news.nd.edu (Tom laughner) (07/12/90)
From article <1990Jul11.161138.13630@dvinci.usask.ca>, by lowey@herald.usask.ca (Kevin Lowey): > From article <1990Jul5.205440.23370@ecn.purdue.edu>, by jmoore@cidmac.ecn.purdue.edu (James D Moore): > But at the same clock speed, the 486 chip appears to > be about twice as fast as the 386 chip (in real mode at least) > > - Kevin Lowey There would be no difference in speed between a 386 with a math coprocessor and a 486. The 486 chip is a 386 + the math coprocessor in one. Intel considers the 486 as a part of the 386 family.
grege@gold.GVG.TEK.COM (Greg Ebert) (07/12/90)
In article <217@news.nd.edu> laughner@news.nd.edu (Tom laughner) writes: >There would be no difference in speed between a 386 with a math >coprocessor and a 486. The 486 chip is a 386 + the math coprocessor in >one. Intel considers the 486 as a part of the 386 family. Actually, the 486 is faster because the instruction cycles were optimized. Obviously, if a 386/25 didn't have a cache, the 486 would run faster. On the flip-side, though, you can build a more efficient cache/cache controller for the 386 than is contained in the 486. Then again, CACHE PERFORMANCE IS A FUNCTION OF HOW THE SOFTWARE WAS WRITTEN. You could write software which would execute with the same speed on systems with 4K, 8K, or maybe 64K of cache. My $0.02 worth: If you have a 386/25 w/80387, don't waste your money on a 486/25. Wait for the 40Mhz version. There are so many other system parameters which limit overall performance. If you do gobs of screen I/O, there will be alsmost zippo difference between a 6Mhz 286 and a 486/33; get a display card which shadows video BIOS. If you think your hard disk is slow, even a 100000Mhz 80786 won't make things faster. If your memory sits on the AT bus, it still runs the same piddly 8Mhz bus cycle. And if TETRIS or BLOCKOUT is too SLOW for you, too bad. A faster CPU won't get those critters to move any faster.
davidsen@antarctica.crd.GE.COM (william E Davidsen) (07/13/90)
In article <217@news.nd.edu>, laughner@news.nd.edu (Tom laughner) writes: |> There would be no difference in speed between a 386 with a math |> coprocessor and a 486. The 486 chip is a 386 + the math coprocessor in |> one. Intel considers the 486 as a part of the 386 family. A popular misconception, but untrue. Let's look at some actual test results which will show that the 486 *at the same speed* is about twice as fast as the 386. These tests were run on 25MHz machines. Note that while integer mpy and divide are about the same, add/sub are faster, bringing the weighted average up. The test and branch, which is a pretty good indicator of performance for non-math stuff such as compilers, shows better than 2:1. Float is about twice as fast. Overall on a 48 program suite I got a2.6 times faster for the 486. System id: HP 486-25, SCO ODT 1.0, 10MB, 300MB, cc Unoptimized math results: Math operations, effective instructions/sec (thousands) Add Sub Mpy Div Wtd. Avg. short: 17347.8 17538.5 3391.3 2845.4 11740.5 long: 19814.7 19600.0 3103.4 1978.0 12897.0 float: 4266.7 4266.7 3824.2 1500.0 3741.1 double: 4133.3 4043.5 3243.2 1360.8 3468.0 int: 19912.1 19384.6 3130.4 2000.0 12871.6 Test and branch timing: integer compare and branch 0.247 uSec, 4054.0K/sec float compare and branch 0.850 uSec, 1176.5K/sec System id: Dell 325, 4MB, 150MB, Xenix/386 2.3.3, 387 Unoptimized math results: Math operations, effective instructions/sec (thousands) Add Sub Mpy Div Wtd. Avg. short: 7088.6 7169.8 2948.7 2631.6 5409.4 long: 7378.6 7169.8 2631.6 1842.1 5298.7 float: 1097.6 1097.6 921.1 897.4 1023.4 double: 963.9 963.9 722.9 786.5 877.0 int: 7378.6 7200.0 2600.0 1818.2 5296.3 Test and branch timing: integer compare and branch 0.688 uSec, 1453.5K/sec float compare and branch 4.320 uSec, 231.5K/sec And if you like some standard benchmarks, Dhrystone 2.1 486: Dhrystone benchmark: Dhrystone Benchmark, Version 2.1 (Language: C) Program compiled without 'register' attribute Please give the number of runs through the benchmark: Execution starts, 600000 runs through Dhrystone Microseconds for one run through Dhrystone: 63.1 Dhrystones per Second: 15838.1 Dhrystone Benchmark, Version 2.1 (Language: C) Program compiled with 'register' attribute Please give the number of runs through the benchmark: Execution starts, 600000 runs through Dhrystone Microseconds for one run through Dhrystone: 61.3 Dhrystones per Second: 16304.3 386: Dhrystone Benchmark, Version 2.1 (Language: C) Program compiled without 'register' attribute Please give the number of runs through the benchmark: Execution starts, 500000 runs through Dhrystone Microseconds for one run through Dhrystone: 126.2 Dhrystones per Second: 7923.9 Dhrystone Benchmark, Version 2.1 (Language: C) Program compiled with 'register' attribute Please give the number of runs through the benchmark: Execution starts, 500000 runs through Dhrystone Microseconds for one run through Dhrystone: 120.0 Dhrystones per Second: 8333.3 Note that these figures match those reported in most of the magazines. The 486 is *not* a 386 with a built-in 387, it is a whole new chip with the same instruction set, but very diferent instruction timings, with the speedup in the common instructions. -- Bill Davidsen (davidsen@crdos1.crd.ge.com, uunet!crdgw1!crdos1!davidsen) GE Corp R&D Center, Schenectady NY Moderator of comp.binaries.ibm.pc and 386users mailing list "Stupidity, like virtue, is it's own reward" -me
psw@richard.mitre.org (Phillip Wherry) (07/13/90)
In article <217@news.nd.edu> laughner@news.nd.edu (Tom laughner) writes: >There would be no difference in speed between a 386 with a math >coprocessor and a 486. The 486 chip is a 386 + the math coprocessor in >one. Intel considers the 486 as a part of the 386 family. This is incorrect. The 486 has an internal cache (8K, I think), and many instructions have been optimized so that they execute in fewer clock cycles. I've got access to both 486/25 and 386/25 machines (with coprocessor in the case of the 386), and I can assure you that there is a LARGE speed difference between the two. Two to two and a half times the speed is about right. -- Phillip Wherry The MITRE Corporation, McLean, VA psw@mitre.org
marshall@wind55.seri.gov (Marshall L. Buhl) (07/13/90)
lowey@herald.usask.ca (Kevin Lowey) writes: >From article <1990Jul5.205440.23370@ecn.purdue.edu>, by jmoore@cidmac.ecn.purdue.edu (James D Moore): >I have a benchmark program that I wrote which does a lot of tests including >the sieve, fibonacci, whetstone, integer math, floating math, >trancendental math, etc. It was done in Turbo Pascal 3.0 so the compiler >does no math optimizations, etc. Here are some of the final results (in >seconds) which I obtained. >Machine Name CPU MHz NO 80x87 With 80x87 >IBM Model 80/486 486 25 26.80 10.92 >DEC 325c 386 25 46.96 >Zenith 386 386 25 51.25 20.93 What do you mean by "With 80x87"? Do you mean compiled for 80x87 or with one installed? I guess you mean the former as the math stuff is built into the 486. -- Marshall L. Buhl, Jr. EMAIL: marshall@seri.gov Senior Computer Engineer VOICE: (303)231-1014 Wind Research Branch 1617 Cole Blvd., Golden, CO 80401-3393 Solar Energy Research Institute Solar - safe energy for a healthy future
pitonyak@dinghy.cis.ohio-state.edu (Andrew Pitonyak) (07/13/90)
>There would be no difference in speed between a 386 with a math >coprocessor and a 486. The 486 chip is a 386 + the math coprocessor in >one. Intel considers the 486 as a part of the 386 family. Not entirely true. The 486 does floating point much faster than the 386 because the FPU is on the chip. When the 386 does a 387 instruction it has to deal with the interface between the two chips, the 486 does not so it is faster with these instructions. By the way, the previously posted timings reflect this. You could improve things a bit by dropping in say a Cyrix chip rather than an intel chip with the 386 machine. I don't know if they have cleaned up and improved the other instructions. I do know that the 33Mhz 386/387 pair is more efficient than the 25Mhz 386/387 combination. Most unusual if you ask me. Andy
6600sirt@ucsbuxa.ucsb.edu (Mike O'Brien) (07/13/90)
From article <217@news.nd.edu>, by laughner@news.nd.edu (Tom laughner): > There would be no difference in speed between a 386 with a math > coprocessor and a 486. The 486 chip is a 386 + the math coprocessor in > one. Intel considers the 486 as a part of the 386 family. You're assuming that just because the 486 appears as a 386+387+cache to software, that it really is a 386+387+cache. In reality, 85% of the surface area of the 486 was dedicated to a RISC processor, meaning that most 486 instructions complete in 1 clock cycle. In addition, the 486 can simultaneously process 5 instructions at once, as long as those instructions don't modify the same registers or memory. (So programmers, start writing programs that do dissimilar things next to each other.) Finally, the 486 can preprocess something like 2048 instructions ahead. (Programmers: write programs that rarely if ever jump more than 2k at a time.) All in all, the 486 ends up being about twice as fast running off-the-shelf software as the 386. But if you write your program with the ^^above^^ programming tips in mind, you can really make it scream. Mike O'Brien 6600sirt@ucsbuxa.ucsb.edu no connection to Intel
lestat@nontech.Berkeley.EDU (David Gonzalez-Nieves) (07/13/90)
>From article <1990Jul11.161138.13630@dvinci.usask.ca>, by lowey@herald.usask.ca (Kevin Lowey): >> From article <1990Jul5.205440.23370@ecn.purdue.edu>, by jmoore@cidmac.ecn.purdue.edu (James D Moore): >> But at the same clock speed, the 486 chip appears to >> be about twice as fast as the 386 chip (in real mode at least) >> >> - Kevin Lowey >There would be no difference in speed between a 386 with a math >coprocessor and a 486. The 486 chip is a 386 + the math coprocessor in >one. Intel considers the 486 as a part of the 386 family. That is not exactly right, the 486 has been designed using newer tecnology than the 386/387. The amount of cycles required for operations in the 486 has been reduced. Also since there is an internal connection between the 486 and it 387 equivalent, it takes less time for floating point operations. The 8K internal cache is also quicker than having an external cache of the same size. The 486 should be quicker than a 386/387/8k SRAM combination running at the same speed. ---------------------------- David Gonzalez Bellcore Summer Intern email: lestat@ctt.bellcore
koziol@yoyodyne.ncsa.uiuc.edu (07/13/90)
One of the main reasons for the '486 being about twice as fast as the '386 is because of the way the chips are clocked. The '386 (and also the '030 BTW) use a divide-by-two clock from the system clock to actually clock the internals of the chip. The '486 (and the '040) are syncronously tied to the system clock, i.e. they actually clock at 25Mhz at a system clock of 25Mhz as opposed to the '386 which is internally dropped to 12.5Mhz at a 25Mhz system clock. I would assume the additional speedups are due to a tighter coupling of the math coprocessor and cache controller, and also some optimization of the silicon. Quincey Koziol koziol@ncsa.uiuc.edu
rdu@cbnewsk.att.com (ranjan.dutta) (07/13/90)
In article <217@news.nd.edu> laughner@news.nd.edu (Tom laughner) writes: >There would be no difference in speed between a 386 with a math >coprocessor and a 486. The 486 chip is a 386 + the math coprocessor in >one. Intel considers the 486 as a part of the 386 family. 486 will be little faster in communicating with the coprocessor as signals do not have to go from one package to onother. Since communication is within the same package, 486-25 may even come out little faster than 386-33 with same amount of memory caching. However, with high cache memory or with low math coprocessor intensive applications, we would be comparing apples and oranges. Ranjan
ggraef@csd4.csd.uwm.edu (gerald graef) (07/13/90)
In article <217@news.nd.edu> laughner@news.nd.edu (Tom laughner) writes: >There would be no difference in speed between a 386 with a math >coprocessor and a 486. The 486 chip is a 386 + the math coprocessor in >one. Intel considers the 486 as a part of the 386 family. The 486 is still faster than a 386+387. This is because many of the support functions are onboard the 486 CPU (like the 8k cache) but are spread over one or more extra chips on a 386. This does make a significant difference in speed. In addition, you can add a numeric processor to a 486 system to get even better speed. -- --Common sense is the collection of prejudices aquired by age 18 - Albert E. --Only by purest chance do the above resemble the views of anyone other than: Gerald Graef: Internet %%%%% ggraef@csd4.csd.uwm.edu : BITNET %%%%% ggraef%csd4.csd.uwm.edu@INTERBIT
garyt@ios.Convergent.COM (Gary Tse) (07/13/90)
laughner@news.nd.edu (Tom laughner) writes: >There would be no difference in speed between a 386 with a math >coprocessor and a 486. The 486 chip is a 386 + the math coprocessor in >one. Intel considers the 486 as a part of the 386 family. Intel claims a CPI of 1.8 for the i486 (this counts internal cache hits, obviously). Glancing at the data book, I don't see any i386 instruction taking less than 2 clocks (a register-to-register move takes 2 clocks). Which is faster? <shrug> It should be obvious. -- Gary Tse, garyt@ios.Convergent.COM || ..!pyramid!ctnews!ios!garyt tse@soda.Berkeley.EDU || ..!ucbvax!soda!tse tse@netcom.UUCP || ..!{amdahl,apple}!netcom!tse "We are errant knaves all; trust none of us."
news@udenva.cair.du.edu (netnews) (07/17/90)
In article <217@news.nd.edu> laughner@news.nd.edu (Tom laughner) writes: >From article <1990Jul11.161138.13630@dvinci.usask.ca>, by lowey@herald.usask.ca (Kevin Lowey): >> From article <1990Jul5.205440.23370@ecn.purdue.edu>, by jmoore@cidmac.ecn.purdue.edu (James D Moore): >> But at the same clock speed, the 486 chip appears to >> be about twice as fast as the 386 chip (in real mode at least) >> >> - Kevin Lowey >There would be no difference in speed between a 386 with a math >coprocessor and a 486. The 486 chip is a 386 + the math coprocessor in >one. obviously, having a 387 built in is much faster. (isn' that obvious?!) it also has a 82385 cache controller, and 8 k of cache. ( i think this is correct, ) my store just got a 486-25 that compiled some autocad pic in 5 min. 53 sec. a 386-33 w/ 387 took 12 min 18 sec. so there. :-) tim
amull@Morgan.COM (Andrew P. Mullhaupt) (07/20/90)
In article <217@news.nd.edu>, laughner@news.nd.edu (Tom laughner) writes: > There would be no difference in speed between a 386 with a math > coprocessor and a 486. The 486 chip is a 386 + the math coprocessor in > one. Intel considers the 486 as a part of the 386 family. Maybe they think of it as a 386 family processor, but the i486 is a hell of a lot faster at the same clock speed. The i486 FPU is on the order of 3 times faster than the 80386/80387 combination at the same clock speed. The i486 is about twice (or maybe a bit faster) than the 80386 at the same clock speed. Also: I have a 25 Mhz i486 which beats up Sparcstation-I by about 30% on identical code. This means that the i486 probably can compete with the Sparcstation-1+ for integer code. It isn't much slower for double precision floating point, but the SS goes faster for single precision (although I have very little use for singles). Later, Andrew Mullhaupt