I've got one of these, Thermaltake A1998: http://www.ebay.com/itm/Thermaltake-A1998-Cop … =item4d03c5c5ad

It's very cheap, and you just add your own 80mm fan. Unfortunately I haven't gotten to use it much as the Voodoo5 build it was going in has had so many problems and isn't running yet, but it is a very impressive heatsink in person. It's all copper, and the fins are really dense; I'm very confident that it will be a great cooler once I get my build working.

One issue I had with it though: the screws that shipped with it (used to install an 80mm fan) were actually the wrong size, so I ended up doing a slightly-ghetto fix by using some thin screws and nuts on the opposite end to hold the fan on. Minor inconvenience for an otherwise quality $8 cooler, but figured I should mention it.

I've got one of those coolers as well. Seems to work OK with an Athlon XP 3200+

The one I have will allow you to reach at least 2.2Ghz (2.5 Ghz here) http://www.thermaltakeusa.com/Product.aspx?S=1135&ID=1554. Its silent and powerful. The only problem is - its big. It wont fit on all mainboards and in all computer cases.

Yeah I was thinking about the typhoon I have access to one. Not sure if the machine i would be using it on has the 754 holes. I was looking at the Ever Cool ND 18 it has a nice copper base with nice tall sharp fins and i was thinking it might not be to bad with with bigger fan and one of those funnel looking adapters.

Then again i was looking at a ton of coolers the other night.

So if a 2500+ is capable of 2.2 or even a massive 2.5 then i guess they were decent clockers no? I put presscot under water and got okay clocks and paired it with 3.5gb ram and an HD2600 and it runs rather well fora P4, I can only imagine a Barton core that is clocked.

cdoublejj wrote:

So if a 2500+ is capable of 2.2 or even a massive 2.5 then i guess they were decent clockers no?

They were fantastic clockers! Especially the 2500+ version which almost certainly could be overclocked to 3200+ (2.2Ghz) just by increasing the FSB. Even better, the mobile (-M) version could reach up to 2.7-2.8Ghz with some luck and some decent mainboard/PSU and cooling.

My barton-m at 2.5 Ghz is pretty limited by the mainboard (I'm using a soft-modded Abit KT7-a, KT133a chipset which doesnt officially support anything newer then Athlon XP 2000+). If I'd want to max out the CPU potential, I'd go for Nforce 2 ultra from DFI or Abit.

Barton AXP-M chips are so cheap now. I like these chips mostly for their low temps. You can usually get a nice overclock while still keeping them under 1.65v. Cranking them to the edge with extra voltage, making them run 50% hotter, seems pointless now though.

Apparently Barton core doesn't even have SSE 2 or 3?

Hi

Here is the CPU info for a BARTON (there is no SSE2/3):

> cat /proc/cpuinfo

1processor       : 0
2vendor_id       : AuthenticAMD
3cpu family      : 6
4model           : 10
5model name      : AMD Athlon(tm) XP 3000+
6stepping        : 0
7cpu MHz         : 2171.656
8cache size      : 512 KB
9fdiv_bug        : no
10hlt_bug         : no
11f00f_bug        : no
12coma_bug        : no
13fpu             : yes
14fpu_exception   : yes
15cpuid level     : 1
16wp              : yes
17flags           : fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 mmx fxsr sse syscall mmxext 3dnowext 3dnow up
18bogomips        : 4343.31
19clflush size    : 32
20cache_alignment : 32
21address sizes   : 34 bits physical, 32 bits virtual
22power management: ts

> cpuid

1 eax in    eax      ebx      ecx      edx
200000000 00000001 68747541 444d4163 69746e65
300000001 000006a0 00000000 00000000 0383fbff
480000000 80000008 68747541 444d4163 69746e65
580000001 000007a0 00000000 00000000 c1c3fbff
680000002 20444d41 6c687441 74286e6f 5820296d
780000003 30332050 002b3030 00000000 00000000
880000004 00000000 00000000 00000000 00000000
980000005 0408ff08 ff20ff10 40020140 40020140
1080000006 00000000 41004100 02008140 00000000
1180000007 00000000 00000000 00000000 00000001
1280000008 00002022 00000000 00000000 00000000
13
14Vendor ID: "AuthenticAMD"; CPUID level 1
15
16AMD-specific functions
17Version 000006a0:
18Family: 6 Model: 10 [Duron/Athlon model 10]
19
20Standard feature flags 0383fbff:
21Floating Point Unit
22Virtual Mode Extensions
23Debugging Extensions
24Page Size Extensions
25Time Stamp Counter (with RDTSC and CR4 disable bit)
26Model Specific Registers with RDMSR & WRMSR
27PAE - Page Address Extensions
28Machine Check Exception
29COMPXCHG8B Instruction
30APIC
31SYSCALL/SYSRET or SYSENTER/SYSEXIT instructions
32MTRR - Memory Type Range Registers
33Global paging extension
34Machine Check Architecture
35Conditional Move Instruction
36PAT - Page Attribute Table
37PSE-36 - Page Size Extensions
38MMX instructions
39FXSAVE/FXRSTOR
4025 - reserved
41Generation: 7 Model: 10
42Extended feature flags c1c3fbff:
43Floating Point Unit
44Virtual Mode Extensions
45Debugging Extensions
46Page Size Extensions
47Time Stamp Counter (with RDTSC and CR4 disable bit)
48Model Specific Registers with RDMSR & WRMSR
49PAE - Page Address Extensions
50Machine Check Exception
51COMPXCHG8B Instruction
52APIC
53SYSCALL/SYSRET or SYSENTER/SYSEXIT instructions
54MTRR - Memory Type Range Registers
55Global paging extension
56Machine Check Architecture
57Conditional Move Instruction
58PAT - Page Attribute Table
59PSE-36 - Page Size Extensions
60AMD MMX Instruction Extensions

61MMX instructions
62FXSAVE/FXRSTOR
633DNow! Instruction Extensions
643DNow instructions
65
66Processor name string: AMD Athlon(tm) XP 3000+
67L1 Cache Information:
682/4-MB Pages:
69   Data TLB: associativity 4-way #entries 8
70   Instruction TLB: associativity 255-way #entries 8
714-KB Pages:
72   Data TLB: associativity 255-way #entries 32
73   Instruction TLB: associativity 255-way #entries 16
74L1 Data cache:
75   size 64 KB associativity 2-way lines per tag 1 line size 64
76L1 Instruction cache:
77   size 64 KB associativity 2-way lines per tag 1 line size 64
78
79L2 Cache Information:
802/4-MB Pages:
81   Data TLB: associativity L2 off #entries 0
82   Instruction TLB: associativity L2 off #entries 0
834-KB Pages:
84   Data TLB: associativity Direct mapped #entries 0
85   Instruction TLB: associativity Direct mapped #entries 0
86   size 2 KB associativity L2 off lines per tag 129 line size 64
87
88Advanced Power Management Feature Flags
89Has temperature sensing diode
90Maximum linear address: 32; maximum phys address 34

Bye bye.

Athlon XP was actually the first AMD chip with full SSE support and there are none with SSE2. The only major disadvantage to this to my knowledge is with H.264/VC-1 hardware decode requiring SSE2. If you want that on your "retro" rig, then go Athlon 64 or P4.

Well some time ago i had put windows 7 on a barton core machine paired with an nivida 6200 and not only did it run aero it was running flash games at the same time and smoothly something that would normally murder a CPU form that era. So i guess it looks like the later P4s have the upper hand.

As said they are great clockers so i'm sure with tweaking it would be just fine even stock clocks it's still fairly snappy.

I'm still curious as to what the copper shims were used for, i was thinking to help prevent cracking of the die?

cdoublejj wrote:

I'm still curious as to what the copper shims were used for, i was thinking to help prevent cracking of the die?

I heard that rumor too - but I never cracked any dies so far and I changed a whole lot of them so I never used them myself. Never really saw need for that. Also, the rumor that it helps cooling down smells like a blatant lie to me. Maybe I'm wrong though... I guess it would be worth finding out, but - meh... 😖

I don't see how it could cool either.I could see how it could possibly keep from cracking dies. My buddy was tech at the shop i work at long before me told me about they saw all kinds of cracked dies.

cdoublejj wrote:

I don't see how it could cool either.I could see how it could possibly keep from cracking dies. My buddy was tech at the shop i work at long before me told me about they saw all kinds of cracked dies.

as a member of the overclocking "scene". I never once saw anyone of my peers crack a core. I've seen many get chipped, but they were doing things you shouldn't be doing. That said, anyone putting a big heatsink on a board almost always used a cpu shim. For $2. Nobody risked it.

ebay:
http://www.ebay.com/itm/280573921428?ssPageNa … 984.m1423.l2649

luckybob wrote:

as a member of the overclocking "scene". I never once saw anyone of my peers crack a core. I've seen many get chipped, but the […]
Show full quote

cdoublejj wrote:

I don't see how it could cool either.I could see how it could possibly keep from cracking dies. My buddy was tech at the shop i work at long before me told me about they saw all kinds of cracked dies.

as a member of the overclocking "scene". I never once saw anyone of my peers crack a core. I've seen many get chipped, but they were doing things you shouldn't be doing. That said, anyone putting a big heatsink on a board almost always used a cpu shim. For $2. Nobody risked it.

ebay:
http://www.ebay.com/itm/280573921428?ssPageNa … 984.m1423.l2649

If Thermaltake Big Typhoon that I use qualifies as such, then I dare to digress 😀 Never used those, and I've used a lot of really big heatsinks 😀

Pfft. It's not a real Socket A cooler unless it makes your ears bleed.

😜

a challenger appears!

http://www.dansdata.com/quickshot011.htm

its load as HELL. I used to have one.

That looks exactly like the cooler I have, paired with a monster fan. Seems like Thermaltake sold that design a couple of different ways.

What in the...that Thermaltake HSF pic looks like a Golden Orb, except it's significantly taller to accommodate the fan, and the lower part's blue.

Just how many variants of that Golden Orb design did they produce?

My Athlon XP 3200+ certainly deserves a better HSF than the stock one...just can't decide on what to get.

That one's a Dragon Orb actually. The xyz Orb range was pretty big. 😀