sandy bridge was insane

[dight]

when i was using a z68 motherboard and a 2500k i could push my cpu to 4.6 ghz but it would only be stable at 4.4 ghz. what gets me is that i now have an r5 3600 and my pc literally bsods at 4.0. it's hilarious a cpu that an 11 year old cpu ocs to a higher clock speed than a zen 2 cpu like this. i mean i am using a shitty b450 board but still. and yet the r5 is like double the speed as the 2500k even when the sandy bridge cpu is oced that much. damn amd really smashed ipc out of the park with zen 2.

[Godlygamer23]

Sandy Bridge was a big deal when it launched. And the fact that it's still kinda usable even now, is kind of impressive. Since Intel was keeping its core count on its consumer CPUs at four, it makes me wonder where we would still be if AMD never stepped up to the plate to create a competitive CPU architecture against Intel. AMD is what caused Intel to create the CPUs it now has, including increased core counts on consumer CPUs. 

[freeagent]

I went from a 3770K that I ran at 4600 to a 3600XT that I ran at 4500. The difference was night and day lol.

[Godlygamer23]

Just now, freeagent said:

I went from a 3770K that I ran at 4600 to a 3600XT that I ran at 4500. The difference was night and day lol.

I had a 3570K that I used since the day I purchased it, back in 2012. It was running at 4.2GHz, and it would run Forza Horizon decently well, but if I drove a car that moved pretty quickly(280MPH for example), the CPU would occasionally not be able to keep up, and I would get a loading screen, and the car would just instantly stop lol. Sometimes the roads wouldn't render in either, so I would drive on nothing for 10 seconds or so. All that went away when I upgraded to my 2600X back in 2019. 

[jaslion]

Different times.

 

Back then the variances in chip quality could be garguantuan so chips ran slower so a higher yield could be had. This then in turn often mean that the k series binned chips could oc a ton.

 

Nowadays most chips are pushed to the limit and only outliers can push far.

 

The baseline has simply moved and thus the performance too out of the box too. We are guaranteed a lot more out of the box than before.

 

Sandy bridgr remained relevant for so long too because intel basically didnt reallt improve more than 20% over the span of 6 years. A leap we are now seeing pretty much every 2 years again. With a simple bog standard oc a 2600k beat a stock 6700k.

 

The moment amd put themselves back into the spotlight things changed massively and pure quad core i5's became irrelevant in the span of little over a year, the highest end 4c/8t i7's became budget tier cpu's and even the early intel answer cpu's like the i5 purr 6cores only really did well for a year or 2 before entering irrelevance too.

 

We are now back in the cpu race and the gpu race is just starting.

[Sarra]

I had a DDR memory kit, way back in the 2000's, that I could get to run stable at more than doube it's rated speed.

 

There were Celerons back in the day that could OC to insane levels.

 

But compared to a AM4 or 12/13th gen Intel platform, the difference in IPC, memory bandwidth, and multi-core performance is night and day. Modern CPU's are binned so carefully, or locked down so tightly, that there's really not much OC available anymore. No headroom. I mean, unless you want to get into XOC.

 

My Ryzen 9 5950X paired with a RX 6800XT pulls less power from the wall at idle than my Haswell based Xeon E3 1280V3 with a GTX 980, also at idle. Pushing a render project through the R9 system, hammering the CPU and GPU at the same time, I can pull more power from the wall than the Xeon system in the same state, but it will finish the project a hell of a lot faster. Render time vs wattage, it's way more effecient to use the R9 for a long project. There's a lot more to this than just 'I used to be able to run a 60% OC and now I'm lucky to get 3%'...

[porina]

45 minutes ago, dight said:

when i was using a z68 motherboard and a 2500k i could push my cpu to 4.6 ghz but it would only be stable at 4.4 ghz. what gets me is that i now have an r5 3600 and my pc literally bsods at 4.0. it's hilarious a cpu that an 11 year old cpu ocs to a higher clock speed than a zen 2 cpu like this. i mean i am using a shitty b450 board but still. and yet the r5 is like double the speed as the 2500k even when the sandy bridge cpu is oced that much. damn amd really smashed ipc out of the park with zen 2.

You're mixing various things here. How are you comparing them? Zen 2 was the first AMD architecture to overall beat Skylake, but if you are looking at multi-thread you got 2 extra cores and SMT on top of that. Without considering clock or IPC that's a 2x advantage over the i5 already (50% extra cores, stack average 30% on top of that for SMT assuming Cinebench). For Cinebench R15, I found my old numbers. Skylake with HT would be about 55% more IPC than Sandy Bridge without. I don't have Zen 2 numbers on hand but it was slightly higher than Skylake from memory.

[Somerandomtechyboi]

50 minutes ago, dight said:

when i was using a z68 motherboard and a 2500k i could push my cpu to 4.6 ghz but it would only be stable at 4.4 ghz. what gets me is that i now have an r5 3600 and my pc literally bsods at 4.0. it's hilarious a cpu that an 11 year old cpu ocs to a higher clock speed than a zen 2 cpu like this. i mean i am using a shitty b450 board but still. and yet the r5 is like double the speed as the 2500k even when the sandy bridge cpu is oced that much. damn amd really smashed ipc out of the park with zen 2.

You just need more voltage assuming you got the cooling if you wanna push it further, apparently 2500k and 2600k are known to do 5g if you got the cooling and arent scared with voltage

 

But like cheddar mill p4 and celeron d also clock really high, probs higher than those sandy cpus

Heres a crappy bios run on air at ~1.8v with 2.4v vpll with my p4 631 thats now dead cause i attempted delid and the die is chipped, yep degraded problably due to vpll, did a windows run abit later on and iirc did 4.8 with 1.5v but only bench stable ht on, e8400 did 4.7 1.55v bench stable, difference is cheddar mills keeps on scaling where wolfdale just hits a brick wall with diminishing returns. I expect a celeron d 347 should clock higher than this, no ht and less cache so should run cooler

 

Thing is sandy clocked high and kept good ipc, netburst clocked ludicrously high but no ipc and dumping power go brrrr

[freeagent]

4 minutes ago, Somerandomtechyboi said:

But like cheddar mill p4 and celeron d also clock really high

 

You mean Cedar Mill right?

 

That other one would be pretty smelly if you ask me 

[Somerandomtechyboi]

3 minutes ago, freeagent said:

 

You mean Cedar Mill right?

 

That other one would be pretty smelly if you ask me 

i mean people called prescott presshot so why not cheddar mill ¯\_ (ツ) _/¯