VOGONS


478 vs 775 insurrection

Topic actions

Reply 80 of 88, by ediflorianUS

User metadata
Rank Member
Rank
Member
2mg wrote on 2023-07-09, 20:14:

Slight necro, but can someone ELI5 how did pre-C2D Intels with 2 "cpus glued together" actually handle FSB?
Did they share it, aka could one "core"choke the other, or was is 1/2 FSB speed split per "core"?

they where not glued together , 1 PCB , 2 dies (worked like dual Xeon 604)

3d26b232e5.jpg
?u=https%3A%2F%2Ftse1.mm.bing.net%2Fth%3Fid%3DOIP.ALbCoJO_oZymlkY3bekfZgHaDi%26pid%3DApi&f=1&ipt=021d88c86a2eaf8e25fb3635c43b47e6a6b979bf7c1057c759ddd5514ee428bc&ipo=images

https://cpumuseum.jimdofree.com/museum/intel/pentium-d/

https://www.bit-tech.net/reviews/tech/cpus/in … 4_670_pd_820/1/

and BTW , under early winXP versions , early vista and even some early 32 bit win7 , in day to day bassis Pentium D series , based on Netburst would handle better than C2D , not to mention posiblity of HT.

My 80486-S i66 Project

Reply 82 of 88, by The Serpent Rider

User metadata
Rank l33t++
Rank
l33t++
ediflorianUS wrote on 2023-07-11, 10:21:

they where not glued together , 1 PCB , 2 dies (worked like dual Xeon 604)

90nm Pentium D were actually glued together. Your second picture is a die shot of the Smithfield core. It also has additional data bus to poke neighbors L2 cache.

I must be some kind of standard: the anonymous gangbanger of the 21st century.

Reply 83 of 88, by 2mg

User metadata
Rank Member
Rank
Member
The Serpent Rider wrote on 2023-07-13, 01:16:
ediflorianUS wrote on 2023-07-11, 10:21:

they where not glued together , 1 PCB , 2 dies (worked like dual Xeon 604)

90nm Pentium D were actually glued together. Your second picture is a die shot of the Smithfield core. It also has additional data bus to poke neighbors L2 cache.

I'm still a bit confused how'd these use the shared FSB, but that explanation might be a bit too technical.

Last edited by 2mg on 2023-07-13, 11:59. Edited 1 time in total.

Reply 84 of 88, by The Serpent Rider

User metadata
Rank l33t++
Rank
l33t++

Mostly as a classical dual socket system, as mentioned before.

I must be some kind of standard: the anonymous gangbanger of the 21st century.

Reply 85 of 88, by TrashPanda

User metadata
Rank l33t
Rank
l33t
The Serpent Rider wrote on 2023-07-13, 01:16:
ediflorianUS wrote on 2023-07-11, 10:21:

they where not glued together , 1 PCB , 2 dies (worked like dual Xeon 604)

90nm Pentium D were actually glued together. Your second picture is a die shot of the Smithfield core. It also has additional data bus to poke neighbors L2 cache.

Intel bodge job CPUs, ran so hot you could cook steak on them!

Fun times for weak VRMs too.

Reply 86 of 88, by ediflorianUS

User metadata
Rank Member
Rank
Member
2mg wrote on 2023-07-12, 23:06:
ediflorianUS wrote on 2023-07-11, 10:21:

Thanks.

By early, do you mean "before they got Service Packs and better support" or "early programs didn't run as good per se"?

before service packs , sometimes even after. PreVista (longohorn editions or beta + SP1)

"Intel bodge job CPUs, ran so hot you could cook steak on them!

Fun times for weak VRMs too." - nonsense TrashPanda , it was even worse with the early lga P4 3.6+ Ghz editions
I had no issues with any D series on Dell cooling. (for 15+ years now)

My 80486-S i66 Project

Reply 87 of 88, by TrashPanda

User metadata
Rank l33t
Rank
l33t
ediflorianUS wrote on 2023-07-14, 12:25:
before service packs , sometimes even after. PreVista (longohorn editions or beta + SP1) […]
Show full quote
2mg wrote on 2023-07-12, 23:06:
ediflorianUS wrote on 2023-07-11, 10:21:

Thanks.

By early, do you mean "before they got Service Packs and better support" or "early programs didn't run as good per se"?

before service packs , sometimes even after. PreVista (longohorn editions or beta + SP1)

"Intel bodge job CPUs, ran so hot you could cook steak on them!

Fun times for weak VRMs too." - nonsense TrashPanda , it was even worse with the early lga P4 3.6+ Ghz editions
I had no issues with any D series on Dell cooling. (for 15+ years now)

You could indeed cook steak on them .. even seen videos of it being done, and weak VRMs did have issues with the Pentium D CPUs. Just because you have had one good experience with one doesn't mean the large body of evidence contrary to your experience is wrong.

Spent way too many hours repairing machines to just take one or even a handful of good experiences as fact. I will admit that in general Dell did have decent cooling in their business machines but they learned from their prior mistakes so their cooling was also generally overkill.

Pentium D was shite - change my mind ..wait dont bother you cant. (Conroe Pentium already did that)

Reply 88 of 88, by 2mg

User metadata
Rank Member
Rank
Member
TrashPanda wrote on 2023-07-14, 12:47:

You could indeed cook steak on them .. even seen videos of it being done, and weak VRMs did have issues with the Pentium D CPUs.

Would a modern cooler ala Scythe Mugen 1/2 s478 handle these monsters?
And maybe some ghetto small coolers and/or small heatsinks to chill VRMs on mobos too?