How much performance benefit is there with pipelining and under which conditions will the benefit be noticed?

Not sure. I sold my DRx2 in 2009, and I didn't really test this feature well.

Anonymous Coward wrote:

If you're not being sarcastic, please provide something to back that up. Any information I've ever been able to find indicates that IBM took a 386 and added cache and 486 instruction set.

Is the Intel 486 really much beyond that?

From wikipedia:

Differences between i386 and i486: (I would imagine listed in the order of importance)
-An 8 KB on-chip (level 1) SRAM cache stores the most recently used instructions and data (16 KB and/or write-back on some later models). The 386 had no such internal cache but supported a slower off-chip cache (which was not a level 2 cache, as there was no internal level 1 cache on the 80386).
-Tightly coupled pipelining allows the 486 to complete a simple instruction like ALU reg,reg or ALU reg,im every clock cycle (even though the latency was several cycles). The 386 needed two clock cycles for this.
-Integrated FPU (disabled or absent in SX models) with a dedicated local bus; together with faster algorithms on more extensive hardware than in the i387, this gives faster floating point calculations compared to the i386+i387 combination.
-Improved MMU performance.

The 386 bus is also slow. The fastest 486 motherboard can deliver over 40mb/sec on the primary DRAM, while the 386 is lucky to get 20mb/sec, or in the case of the 386sx 10mb/sec.

The way I see it, if the motherboard uses a 386SX or 386DX bus, the system is still a 386 regardless of what CPU is installed.

What if you use a 486DLC with L1 disabled and never run any software requiring the 486 instruction set?

Motherboard is Aquaris MB-4D33/50 OPTi 495SX Motherboard.

My Dataexpert OPTI 495SLC motherboard is broken because Hardcore testing.

Not use any software/drivers. (such as 486drx utility)

Motherboard is Aquaris MB-4D33/50 OPTI 495SX Motherboard.

Not use any software/drivers. (such as 486drx utility)

Motherboard is Aquaris MB-4D33/50 OPTI 495SX Motherboard.

same 40MHz, same FPU, but fpu performance reduced 50%.

very interesting result.

Motherboard is Aquaris MB-4D33/50 OPTI 495SX Motherboard.

reason of upload this result is refernce of motherboard's performance.

Speedsys rating of 9.04 for the DRx2 33/66 vs 8.84 for the DLC-40 is very close (TI486SXL gets 10.13). This is probably the reason why I sold my DRx2 in the first place, the clock doubling seems to offer very little. The DRx2 is so rare and expensive, and the DLC-40 so common and cheap. Unless you are a collector, why bother?
Actually, the DRx2 is supposed to have built in cache coherency which makes it desirable for people who want to upgrade very old 386s.

The increase in FPU performance is interesting. I noticed the jump using Cyrix chips as well. I wonder if it is real, or just a benchmark anomaly.

I like the Aquarius motherboard. I used to have the same one. It has very good compatibility with a wide range of chips...it even allowed my Adaptec 1542C and DLC to work in the same system. I assume this board has the 495SX chipset. You should run the doom benchmarks again. I bet it's slower than the other board.

Results from my IBM PS/2 Model 70. This one has the IBMish 20 MHz 386 with no cache (Like this one).

Full report (including a report for the 120 MB DBA drive):

1 ───────────────────────────────────────────────────────────────────────────
2System Speed Test  Ver 4.78      Report file - created on 20-01-2016 20:56:11
3 ───────────────────────────────────────────────────────────────────────────
4
5CPU is in V86 mode: No
6Processor         : 80386DX   21 MHz
7CPU speed index   : 2.29
8
9Total memory size : 6 MB
10Memory Bandwidth  : 47.17 MB/s
11VESA video memory : Unknown
12
13Hard drive 0      : 115C 64H 32S 115 MB
14                  Tested in FULL mode
15                  Average/Max seek time  : 26.38 ms / 40.88 ms
16                  Random seek time       : 23.44 ms
17                  Track-to-track seek    : 10.15 ms
18                  Random access time     : 35.14 ms
19                  Linear verify speed    : 552 KB/s
20                  Min/Max verify speed   : 281 KB/s / 588 KB/s
21                  Linear read speed      : 439 KB/s
22                  Min/Max read speed     : 318 KB/s / 579 KB/s
23                  Buffered read speed    : 766 KB/s
24                  Hard Drive speed index : 5.44
25
26Year2000 Bug      : 31-12-1999  01-01-1900
27OS version        : MS-DOS 7.00
28 ───────────────────────────────────────────────────────────────────────────
29
30                ATA/ATAPI Device Information
31
32IDE0─┬─Master─<Empty>
33     │
34     └─Slave──<Empty>
35
36IDE1─┬─Master─<Empty>
37     │
38     └─Slave──<Empty>
39
40IDE2─┬─Master─<Empty>
41     │
42     └─Slave──<Empty>
43
44IDE3─┬─Master─<Empty>
45     │
46     └─Slave──<Empty>
47 ───────────────────────────────────────────────────────────────────────────
48
49                Cache/Memory Benchmark
50                ┌──────────────┬──────────────┬──────────────┬──────────────┐
51                │     Read     │     Write    │     Move     │    Average   │
52┌───────────────┼──────────────┼──────────────┼──────────────┼──────────────┤
53│ Memory        │    10.41 MB/s│    14.72 MB/s│    14.45 MB/s│    13.19 MB/s│ 
54└───────────────┴──────────────┴──────────────┴──────────────┴──────────────┘
55 ───────────────────────────────────────────────────────────────────────────
56
57                Memory Modules Information
58
59           SMBus/i2c Host Controller     : Not Present
60 ───────────────────────────────────────────────────────────────────────────

61
62End of report.
63

Anonymous Coward wrote:

Speedsys rating of 9.04 for the DRx2 33/66 vs 8.84 for the DLC-40 is very close (TI486SXL gets 10.13). This is probably the reas […]
Show full quote

Speedsys rating of 9.04 for the DRx2 33/66 vs 8.84 for the DLC-40 is very close (TI486SXL gets 10.13). This is probably the reason why I sold my DRx2 in the first place, the clock doubling seems to offer very little. The DRx2 is so rare and expensive, and the DLC-40 so common and cheap. Unless you are a collector, why bother?
Actually, the DRx2 is supposed to have built in cache coherency which makes it desirable for people who want to upgrade very old 386s.

The increase in FPU performance is interesting. I noticed the jump using Cyrix chips as well. I wonder if it is real, or just a benchmark anomaly.

I like the Aquarius motherboard. I used to have the same one. It has very good compatibility with a wide range of chips...it even allowed my Adaptec 1542C and DLC to work in the same system. I assume this board has the 495SX chipset. You should run the doom benchmarks again. I bet it's slower than the other board.

i think 486DLC 40MHz,486DRx2 66MHz,486SXL 40MHz have diffrent goods and bads.

486DLC 40MHz - very ceap and common. performance is not bad. good cache/memory performance because 40mhz fsb
486SRX 40MHz - uncommon and expensive. performance is very good. 8KB L1 Cache made this cpu real 486. good cache/memory performance because 40mhz fsb
486DRx2 66MHz - uncommon and expensive. performance is good. can use 66mhz 387 fpu. good vlb card compatibility because 33mhz fsb.

CPU : AMD 386SX 33MHz
FPU : Cyrix FasMath CX83S87-33-GP 33MHz
M/B : Seritech ALI1217 386SX Motherboard
RAM : 70ns 16MB (4MB x 4)
VGA : ATI VGA Wonder 1024D ISA
I/O : IWill SIDE ISA+I/O
HDD : IBM 12GB 2.5" IDE

speedsys benchmark is here : https://youtu.be/EpH2NLcifh0
sandra benchmark is here : https://youtu.be/GRIiRui9DIE

CPU : AMD 386SX 33MHz
FPU : ULSI SX/SLC US83S87 40@33MHz
M/B : Seritech ALI1217 386SX Motherboard
RAM : 70ns 16MB (4MB x 4)
VGA : ATI VGA Wonder 1024D ISA
I/O : IWill SIDE ISA+I/O
HDD : IBM 12GB 2.5" IDE

speedsys benchmark is here : https://youtu.be/_cj_b6W1UU4
sandra benchmark is here : https://youtu.be/xvDWmiXxYJw

CPU : AMD 386SX 33MHz
FPU : IIT 3C87SX-33 33MHz
M/B : Seritech ALI1217 386SX Motherboard
RAM : 70ns 16MB (4MB x 4)
VGA : ATI VGA Wonder 1024D ISA
I/O : IWill SIDE ISA+I/O
HDD : IBM 12GB 2.5" IDE

speedsys benchmark is here : https://youtu.be/TFaNDdQqEDk
sandra benchmark is here : https://youtu.be/0O9tJif1L6E

You guys are posting pics of really fast 386s. A SX/20 is missing I guess, so here's my Samsung Deskmaster 386SX@20MHz just for reference. No cache.
(Yes, it's an SX wrongly recognized by SpeedSys as DX).

Skv400, I can't speak for the rarity of the ti486sxl, but it is certainly far less expensive than the 486drx2 33/66, which can fetch over 100usd. I have 3 ti486sxl and paid no more than $15.

Konc, are you certain your system has no cache? Your speedsys graph almost looks as if your chipset has 16kb cache.

Anonymous Coward wrote:

Skv400, I can't speak for the rarity of the ti486sxl, but it is certainly far less expensive than the 486drx2 33/66, which can fetch over 100usd. I have 3 ti486sxl and paid no more than $15.

Konc, are you certain your system has no cache? Your speedsys graph almost looks as if your chipset has 16kb cache.

really?i can't search 486sxl. only 486sxl2 on ebay germany seller(amoretro) at $70 price.

i have 486sxl2 50mhz but it is 168pin cpu.

and i don't know 386sx system's memory performance better than opti 495slc/sx based 386dx motherboards.

opti 495's memory performance is not to good?

Ask if anyone on CPU world has a spare, that's how I got one of mine. The sxl2 chips on eBay are overpriced. The seller implies they are make it 486 chips, which they are not because they lack the cache coherency circuit.

Both sxl 40s and sxl2 50s default to 1x clock mode, and both can enable clock doubling by toggling the direct mapped/2 way associated cache register in the Cyrix dlc software util. Both chips are essentially the same exept for the silkscreen label.

386sx motherboards certainly do not have better memory performance than DX chips, they should be half the speed. Opti does indeed have slow main memory performance, but they work okay and are common.

Anonymous Coward wrote:

Konc, are you certain your system has no cache? Your speedsys graph almost looks as if your chipset has 16kb cache.

Positive, I was referring to L2 cache on the m/b. 16KBs of on-chip L1 are obviously there and correctly detected.
In fact I was puzzled by SpeedSys's incorrect detection of this SX. I did open the case and visually confirmed it's an SX@20MHz and, just to be 100% sure, also ran a couple of other "system information" utilities that did identify it as such.