When did you realize that AMD FX processors are pretty good? A lot of the bad press was because games were still largely single threaded
when I switched to intel and got triple the performance without having to do ridiculous overclocks and waste electricity on inferior AMJeet hardware
>>109904764When I replaced my Phenom II X4 with an FX-8350
>>109904764When I needed virtualization features
>>109904764i have one sitting right now with no mobo should I get a mobo for it
>>109905043it's nearly winter you can use it as a heater
>>109905093keki figured
>>109905093The Bulldozer+Crossfire R9 295X2 space heater seems almost cute when you compare it to modern Intel+Nvidia systems.
>>109904764Still got an FX-6300 sleeper in a box somewhere.What a shame we never got Steamroller and Excavator (3rd and 4th gen Bulldozer architecture for those not knowing) full AM3+ FX CPUs. We only ever got the 1st and 2nd gen but 3rd and 4th only got used in cut down low power lower core count chips.It was probably decided like that to make Ryzen shine brighter on its release. It wouldn't look as good compared to a 4th gen FX rather than 2nd gen FX as it was in reality.
>>109905117I've used my FX + Vega on the winter and I don't think it's an effective heater
What was the issue with these again? Each "core" was a stretch of the definition because apparently a core should have a seperate ALU and FPU and these FX cores shared an FPU which bogged down float heavy workloads? I get the gist of it?
>>109905917Yeah that's the gist of it. There was some more shared stuff but the FPU was the major problem - sharing other components between cores is common even in today's CPUs. Some inteltards compared it to HT but HT is just some extra registers to hold data for the thread. It was much more than HT, but still a little bit short of full cores.Later Bulldozer generations like the ones anon >>109905214 mentions improved on things a little, but this major drawback of sharing stuff remained.
>>109905917main issue>shitty 32nm process that even K10 could barely reach 3GHz other issues>huge L2 latency>tiny L1 data cache>write through L1 data cacheshared fpu was rarely an issueshared front end was rarely an issue
>>109905917same issue as 7000 series gpus from amd. the core is good, the board is bad. the board will die or show errors way before the core. a cpu/gpu is only as good as the board its socketed in.
>amd fantards are now retroactively glazing even amd's shitty processors
feeling cute might benchmark the fpu vs shared front end situation
>>109906407amd makes good processors but shitty boards back then and to this day.9800x3d is godlike cpu for gaming but will die in the average asrock/asus motherboard
>>109906408FX 4200 (4M4C) is not faster than FX 4100 (2M4C)
>>109906435I saw that youtube videoi'm thinking a couple dhrystone and whetstone runs to see how it scalesI'll be testing an excavator cpu
https://www.youtube.com/watch?v=8zTzpYjQ2MM
>>109905214>was probably decided like that to make Ryzen shine brighter on its release.It had nothing to do with Ryzen, they just simply didn't have the resources. If they made high-end steamroller/excavator chips, they'd sell at very low margins if not at a loss, so they'd not recuperate neither the engineering and silicon costs. They were busy making console chips and gearing up for the Ryzen launch.Remember, AMD stock was $2 at the time. They would've been bankrupt if not for the console deals, today they are a trillion dollar company and still can't produce GPUs neither in large numbers nor at very high end. Only reason they even went that high is because Ryzens can specifically run good with small chips glued together (first gen ryzens were monolithic and barely broke 3GHz).
i5 2500k destroyed amd because it could hit 5.4ghz. amd was lucky to hit 4ghz boost, on average it was more like 3.6ghz.
>>109906463First gen ryzens had a 14nm chiplet and and a 14nm io die. the apoos were monolothicthey ran 4 Ghz if you were lucky, 3.8GHz was an average oc.
>>109906481actually fx processors were used to break oc records back then, they were worse than equivalent intel at the same ghz though cause they had worse ipc
>>109906497the max temp of amd phenom and fx chips were 62-65c. exceed that and the cpu crashes / mobo trips thermal shutdown. they were way behind intel at the time.
>>109906508you just had a shit mobo
>>109906514https://hardforum.com/threads/temperature-range-for-a-phenom-ii.1510706/>i think 70c is where the core starts to shutdown to protect it selftoday 70c is just casual boost temps.thermal shutdown is more like 95c. 65-70c is like no headroom for oc for amd.
>>109906514No the chips were hardwired to shut down at much lower temps than Intel. See >>109906460
>>109906514Wrong Stupid Faggot
>>109906523>>109906532>>109906536uh oh retarded faggot is having a meltie, im sorry youre retarded i told you fx processors were used to set records and you link me a phenom thread. maybe stop being a moron? haha
>>109904764phenom II blacked edition was better
>>109906490first gen ryzen was monolithic, so no chiplets>>109906481Record on FX 8350 is 8.7GHZ, 2500k could only hit 6.3GHz, so idk why are you comparing heavily overclocked i5 against stock fx>>109906508>>109906523this was issue for all old cpus even intelare you bots, poopjets or poopjet bots??
>>109906621>first gen ryzen was monolithic, so no chipletsYes I looked it up now>this was issue for all old cpus even intelIntel nodes could run much hotter. At one point in the netburst era they removed the temp sensor from the cpu so people wouldn't be scared of 100+C tempsPhenom II 45nm GloFo absolute max temp was 62 C and they got unstable over 55 C when heavily overclockedFX 32nm GloFo ran fine up to 70 C above which you risked degradation. I've seen reports.
>>109906408as you can see even the integer cores don't show perfect 4x scaling when the front end gets clogged upcpu was not throttling the APM power limit is very generous for such a tiny laptop cpu
>>109904764They were still selling this dogshit in “budget” laptops in 2020 long after Ryzen. Slow and painful to use for any modern tasks including Youtube.
>>109906804those are mega cut down chips with extremely small cache at super low clock speedsa full fat 8350 still holds up
>>109904764why would I think or care about antiquated products?
>>109904764I Love my AMD SpaceheaterIt may be Cold OutsideBut I'm Nice & Toasty Inside with my AMD Spaceheater!
>>109906832>a full fat 8350 still holds upI doubt it. I had a 6350 in 2016 and already back then it felt outdated.
>>109904764The problem is that, then as now, AMD was selling them for maybe $100 less than their competitors and no one wanted to buy junk when they could just spend a little more and get an actual good CPU. $100 difference over the 7 years you'll use a PC is basically nothing.If they were being hawked for $50 or something, they'd have made way more sense.
>>109907389it was outdated, definitely. The pcie controller was still in the northbridge.mind the 8350 offered a flat 33% increase in core count and with a minor cpu-nb tweak you could easily make the memory and l3 cache considerably faster
