Blame the laptop maker for using a BGA chip. Intel still does provide socketed chips for mobile.

The only reason AMD tries so hard to keep processors compatible with existing sockets is to provide a good upgrade path. And they try hard to provide an upgrade path to retain their user base. Since their processors draw too much power and offer underwhelming performance, if an existing AMD user were to upgrade and they did not offer that upgradeability, he would have to get a new motherboard as well, and at that point it would be very easy to just switch to Intel as a similar price. But as it is now, it becomes a choice between just buying a chip or a motherboard+processor combo, making staying with AMD a more attractive option.

Intel doesn't do that because they do not feel threatened. That being said its no big loss... yeah you can't upgrade your old Sandy Bridge system to Haswell, but its not like a good i5/i7 from five years ago can be outperformed by the latest and greatest Bulldozer spawn besides some very specific cases. And in the end Intel is much more free to try stuff out on their processors, like adding an integrated voltage regulator (you can't do that and keep the old socket/mobo compatibilioty) to save power, and then removing it one generation later when it doesn't pay off.

washu wrote:

I think you are remembering through the fog of nostalgia. Back when AMD chips were competitive or better than Intel chips they were just as expensive or more so. I had AMD chips in both my personal and professional life when they were the best. I will again if that happens in the future. Today AMDs are not competitive at all in performance or price in the markets I use them in.

It is a possibility, but I remember a 1.8Ghz P4 + shitty SDRAM motherboard cost as much as a 2000+ plus a decent DDR board. It was a decision I had to make back then.

washu wrote:

Oh the irony given the purpose of this thread!

I fail to see the irony. My motherboard's chipset and socket support the FX series CPUs. The ASUS M5A78L-M LE/USB3 , Asrock 760GM-GS3 and Gigabyte GA-78LMT-S2 all support FX series CPUs and share the same socket / chipset as my Biostar. The fault lies with Biostar alone on this one.

washu wrote:

This is obviously not true at all. Lots of laptops have HM70 chipsets with many different CPUs other than Pentiums. One faulty laptop is not indicative of all of them.

Oh really? Read up: https://www.bios-mods.com/forum/Thread-Asus-X55A-30-minutes You do imagine I checked both the TDB and cooling ability of said laptop before sticking an i5 in right? The machine runs flawlesly, but it will shut down in PRECISELY 30 minutes to the millisecond with anything other then a Pentium Dual Core or Celeron CPU.

From a cnet.com article:

"The problem is that Intel has introduced an artificial limitation to handicap the unit by making it turn off after 30 minutes like clockwork. It is a hard shutdown like when a system overheats. People who have done this have monitored temps which were fine and always find it turns off at 30 minutes to the second."

https://www.google.ro/search?q=hm70+cpu+shutd … sm=122&ie=UTF-8

washu wrote:

and if you want to upgrade the CPU in your desktop you more then likely have to spend nearly twice since you need a different MB as well. I have where the IT industry is heading.
Intel chipsets and MB have generally supported two generations per since at least the core 2 days.

Well you can stick an AM3 AMD CPU in an AM2+(and some AM2 boards) and it will run just fine as long as the BIOS supports it. Intel went with LGA1366 for the original i7, the 1156 for the i5 and later 8xx series i7 cpus -> then jumped to LGA 1155 for sandy bridge and ivy bridge - after that you have 1150? for Skylake and lately 1151 for Skywell.

I can understand why they switched from LGA2011 to 2011-2 with the enthusiast class CPUs (DDR4) but LGA 1155 should have been fine for Sandy/Ivy/Hanswell/Broadwell. Skylake platform change is warranted tough.

washu wrote:

It's not like AMD was any better when they were in the lead, they switched sockets and chipsets quickly back when the Athlon 64 came out in the lead.

Yeah, the whole socket 754/939 was pretty annoying. Especially since the were both in use at the same time. It created quite a bit of confusion for some users. Back the I was working for a big hardware reseller and you wouldn't believe how many people bought a socket 939 board with a socket 754 cpu or vice-versa. Just goes to show whoever is on top will turn into an asshole.

alexanrs wrote:

Blame the laptop maker for using a BGA chip. Intel still does provide socketed chips for mobile.

The only reason AMD tries so hard to keep processors compatible with existing sockets is to provide a good upgrade path. And they try hard to provide an upgrade path to retain their user base. Since their processors draw too much power and offer underwhelming performance, if an existing AMD user were to upgrade and they did not offer that upgradeability, he would have to get a new motherboard as well, and at that point it would be very easy to just switch to Intel as a similar price. But as it is now, it becomes a choice between just buying a chip or a motherboard+processor combo, making staying with AMD a more attractive option.

Pretty much yeah.

alexanrs wrote:

Intel doesn't do that because they do not feel threatened. That being said its no big loss... yeah you can't upgrade your old Sandy Bridge system to Haswell, but its not like a good i5/i7 from five years ago can be outperformed by the latest and greatest Bulldozer spawn besides some very specific cases. And in the end Intel is much more free to try stuff out on their processors, like adding an integrated voltage regulator (you can't do that and keep the old socket/mobo compatibilioty) to save power, and then removing it one generation later when it doesn't pay off.

Performance-wise you are correct. It doesn't really make sense to go from sandy to hanswell, but the fact that you require a platform change to do so is what bothers me. Please stop thinking like that. Quick and easy upgrade paths were common in the past, but not now. This needs to change. It might not be important to you, but trust me, it's VERY important to many of us.

Voltage regulator integrated into the CPU? Why? What was wrong with having it on the motherboard instead? Sure, intel did some pretty cool stuff with it's past generations, but integrated voltage regulators, K vs non-K series CPUs + removing the PCI-E bus divider something I don't agree with.

alexanrs wrote:

Intel doesn't do that because they do not feel threatened.

This is exactly why we need healthy competition.

Erm... Haswell introduced an integrated voltage regulator module (on die), which changes how power has to be delivered to the chip. If new motherboards had to be created to support this in the first place, why bother keeping the socket?

alexanrs wrote:

Erm... Haswell introduced an integrated voltage regulator module (on die), which changes how power has to be delivered to the chip. If new motherboards had to be created to support this in the first place, why bother keeping the socket?

Why introduce the integrated VRM? What's wrong with having it on the motherboard in the first place? It's a useless feature in my book - certainly not one to warrant a socket change (from an end-user point of view).

Better efficiency, as Intel was making a big push for energy efficiency with Haswell. This apparently moved another source of heat to the die and did not really pay off, so they removed it.

Right - now related to the original topic - I've managed to correctly add the microcode needed for the FX CPU line but the board still won't post. I get memory beeps like there's no ram in the board, regardless of what speed / voltage / latency ram I try.

There's quite a bit more to do so a board can support FX CPUs, so I'm giving up. I ordered the Asrock 970 mATX board and will get it on the 14th.

Dang sorry to hear that man. Hope the Asrock board works out well for you, every single one I have purchased has been great so far. No angry customers or friends 😀

I gave up after doing more extensive research. It would seem the AGESA microcode in AMIBIOS is somehow embedded into a core module flagged as 1B (single arch bios) - sometimes even 1E, 2E and 3E modules contain microcode fragments. Apparently one has to extract said modules and add specific AGESA components into each one, which people are trying to do via HEX editing. For my board, AGESA code fragments are stored in 1B Single Link Arch Bios module as well as 1E - Reserved.

If the board was running AWARD bios, it could be done since it stores AGESA code in a separate module.

So I finally got the new mainboard in yesterday - an ASRock 970M PRO3 and got the machine running. Now I have the old mainboard and CPU as spares. Here are some pics of how everything came out:

Wire management is almost non-existent at this point, but I'll get around to it when I'm bored. The thing is cooled by 2x120mm quiet fans and one 120mm noisy fan that came with the water cooler (I keep it throttled down to 1000rpm unless the CPU gets over 50 celsius - then it goes to 2000rpm). The lighting is provided by a red LED strip soldered to a 12v molex.

I'd like to thank everyone who offered suggestions. Too bad modding the old board didn't pan out. The bios thing aside, it seems the CPU socket needed to be modded too. The FX series (AM3+) uses a 3.4 MHz VID while AM3 socket can only handle a 400 KHz VID, and my socked doesn't support said VID.

kanecvr wrote:

Yeah, the whole socket 754/939 was pretty annoying. Especially since the were both in use at the same time. It created quite a bit of confusion for some users. Back the I was working for a big hardware reseller and you wouldn't believe how many people bought a socket 939 board with a socket 754 cpu or vice-versa. Just goes to show whoever is on top will turn into an asshole.

I've actually made this exact mistake, buying a s754 board and buying a s939 CPU (or the other way around, don't remember but they wouldn't fit 🤣) so instead of returning the motherboard, I decided to get another board and CPU and build myself 2 A64 rigs 😁

Tetrium wrote:

kanecvr wrote:

Yeah, the whole socket 754/939 was pretty annoying. Especially since the were both in use at the same time. It created quite a bit of confusion for some users. Back the I was working for a big hardware reseller and you wouldn't believe how many people bought a socket 939 board with a socket 754 cpu or vice-versa. Just goes to show whoever is on top will turn into an asshole.

I've actually made this exact mistake, buying a s754 board and buying a s939 CPU (or the other way around, don't remember but they wouldn't fit 🤣) so instead of returning the motherboard, I decided to get another board and CPU and build myself 2 A64 rigs 😁

Even the people working at the company's stores were making this mistake so don't worry about it. I remember at one point we had the store's manager post notices warning clerks not to mix up 939 and 754 parts 😀

Worse, more then once special clients tried (and some kind of managed) to forcibly insert socket 754 CPUs in 939 mainboards, and then brought the whole machine in the shop saying "the new parts are defective it won't work" 😁

kanecvr wrote:

Tetrium wrote:

kanecvr wrote:

Yeah, the whole socket 754/939 was pretty annoying. Especially since the were both in use at the same time. It created quite a bit of confusion for some users. Back the I was working for a big hardware reseller and you wouldn't believe how many people bought a socket 939 board with a socket 754 cpu or vice-versa. Just goes to show whoever is on top will turn into an asshole.

I've actually made this exact mistake, buying a s754 board and buying a s939 CPU (or the other way around, don't remember but they wouldn't fit 🤣) so instead of returning the motherboard, I decided to get another board and CPU and build myself 2 A64 rigs 😁

Even the people working at the company's stores were making this mistake so don't worry about it. I remember at one point we had the store's manager post notices warning clerks not to mix up 939 and 754 parts 😀

Worse, more then once special clients tried (and some kind of managed) to forcibly insert socket 754 CPUs in 939 mainboards, and then brought the whole machine in the shop saying "the new parts are defective it won't work" 😁

I've seen some stuff that's simply unbelievable which for a moment made me lose confidence in humanity 🤣
When I went through a few lots of boxes of spare parts, I found some stuff my boss has handed down to me.

He used to label everything. Now the labeling wasn't the problem, but if one labels a "Pentium 2" cartridge as a "Pentium 3" (and I mean literally a few centimeters away from the P2 logo!) and labels a PCI graphics card as an AGP card (or vice versa, but anyway it was totally wrong...how the hell does one repair computers and not be able to distinguish a PCI adapter from a AGP adapter? That's either impossible or insane 🤣!).

But I knew quite quickly what mistake I made, I simply didn't check the motherboard model number up carefully enough, I just wasn't paying attention.
But in the end both systems worked out perfectly for me! 😁

kanecvr wrote:

I gave up after doing more extensive research. It would seem the AGESA microcode in AMIBIOS is somehow embedded into a core module flagged as 1B (single arch bios) - sometimes even 1E, 2E and 3E modules contain microcode fragments. Apparently one has to extract said modules and add specific AGESA components into each one, which people are trying to do via HEX editing. For my board, AGESA code fragments are stored in 1B Single Link Arch Bios module as well as 1E - Reserved.

If the board was running AWARD bios, it could be done since it stores AGESA code in a separate module.

Really? Dude your unreal if you can help me brother, I'll love you forever & ever.

Can someone help me add simply this 1 microcode to my bios? fx 8100 CPUID 600F12

It's Award BIOS buddy. 😊

Nope. Not doing that here. Go to sites that specialize in this.

Thread closed.