• Competitor rules

    Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.

Ryzen Sandy Bridge IPC

Edit: oh i see ^^^^ ok :)

Haswell i5.

From the top score down.

4.7Ghz: 186
4.6Ghz: 181
4.5Ghz: 175

Its not 24/7 stable @ 4.7Ghz, at least not at anything less than 1.35v.
4.6Ghz is but still at 1.3v, sometimes i run it at that speed for a while.
Mostly 4.5Ghz, 1.26v, nice, cool and quiet.

Haswell according to Anand has 12% higher IPC in Cinebench than Sandy

hbdfh.png
 
All guys are missing CRUCIAL point. Sandy 2500k/2699k did 4900-5200 depends on chip. And anything below that cant be called proper Overclocking.

That 2500k is runnign at 5ghz with HIGH volts if i remember it was like 1.44 Constant volts 7 years now !!! And that got at lest 300 days of constant load as thats my Played time in WoW On it besides other uses :D

I always compare CPUs Max possible overclock vs max possible Overclock.

so what 3770k is faster 4.5 vs 4.5 2600k ?? When sandy can do 5ghz and ivy can do only 4.7 ??And that 5ghz IS Faster than 4.7. Ipc gain is ipc gain but for gaming Clockspeed is also important !!!

my 5820k taht maxed out at 4.45 was SLOWER in games than 2500k@5ghz

But i also do video editing and thats why i went for 6 cores :) And thats why i got 8 cores now :p
 
I think at this point i do agree with AndreiD, Kabylake / Skylake <same thing; IPC is just too much beyond Sandy for it to get anywhere near in performance, even Haswell has more IPC than its overclockable deficit to Sandy, a good chunk more.

And because this is a Ryen vs Sandy IPC thread, in another thread we did establish Ryzen has an Single threaded Cinebench IPC that is 5% higher than Haswell, 8% lower than Kabylake, but because of the differences in SMT performance Ryzen more than makes that up in MT Cinebench performance.

So the answer i guess is Sandy has has a better chance against Ryzen than Kabylake, but its still outclassed by quite a chunk vs Ryzen.

To think a year ago we said we would have been perfectly happy with 6 core Sandy-Zen, its much more than that.
 
And how is sandy at 5Ghz doing in games nowadays on modern titles? All this single thread jazz going on... who actually runs a cpu with just one core? Useful for comparison if you factor in the fact that cores don't operate in isolation.

I am sure there are some video's out there of highly clocked SB's vs clocked other chips in games, and reviews.

Also... SB's didn't ALL clock high. A lot of people were stuck at 4.5ish. Claiming that they nearly all went to 4.8Ghz is BS.

Let's not forget the magic of the solder as well. Everything after SB appeared to overclock worse due to intel changing from solder to thermal paste. SB was great because Intel did the job properly, rather than cheaping out.
 
I think at this point i do agree with AndreiD, Kabylake / Skylake <same thing; IPC is just too much beyond Sandy for it to get anywhere near in performance, even Haswell has more IPC than its overclockable deficit to Sandy, a good chunk more.

And because this is a Ryen vs Sandy IPC thread, in another thread we did establish Ryzen has an Single threaded Cinebench IPC that is 5% higher than Haswell, 8% lower than Kabylake, but because of the differences in SMT performance Ryzen more than makes that up in MT Cinebench performance.

So the answer i guess is Sandy has has a better chance against Ryzen than Kabylake, but its still outclassed by quite a chunk vs Ryzen.

To think a year ago we said we would have been perfectly happy with 6 core Sandy-Zen, its much more than that.

Yeah pretty much. All offer similar performance per core stock or overclocked. What we need to see is a 24/7 stable Ryzen around 4.5Ghz I think that would cover any scenario.
 
Last edited:
And how is sandy at 5Ghz doing in games nowadays on modern titles? All this single thread jazz going on... who actually runs a cpu with just one core? Useful for comparison if you factor in the fact that cores don't operate in isolation.

I am sure there are some video's out there of highly clocked SB's vs clocked other chips in games, and reviews.

Also... SB's didn't ALL clock high. A lot of people were stuck at 4.5ish. Claiming that they nearly all went to 4.8Ghz is BS.

Let's not forget the magic of the solder as well. Everything after SB appeared to overclock worse due to intel changing from solder to thermal paste. SB was great because Intel did the job properly, rather than cheaping out.

You need to make a distinction between the Sandybridge chips. The 2600K and 2700K don't seem to run as high as the 2500K and 2550K. Also overclocking Sandybridge can be a little hit mis as you see failures to boot at certain clocks speeds. Like you find a whole at 4.3Ghz and get a failure to boot, but 4.4 runs fine.

Anyway it's much of a muchness. Sandy is dead now so how well it overclocked or not is irrelevant.
 
Last edited:
You need to make a distinction between the Sandybridge chips. The 2600K and 2700K don't seem to run as high as the 2500K and 2550K. Also overclocking Sandybridge can be a little hit mis as you see failures to boot at certain clocks speeds. Like you find a whole at 4.3Ghz and get a failure to boot, but 4.4 runs fine.

Anyway it's much of a muchness. Sandy is dead now so how well it overclocked or not is irrelevant.

When I say SB I am in general referring to the i5 2500k that started this thread. However.... instructions also got added to later chips 3xxx onwards etc, so the 2600k etc will still be at a disadvantage in certain applications (like my Planetside 2 example) when it comes to actual usage, rather than just a benchmark.
 
And how is sandy at 5Ghz doing in games nowadays on modern titles? All this single thread jazz going on... who actually runs a cpu with just one core? Useful for comparison if you factor in the fact that cores don't operate in isolation.

I am sure there are some video's out there of highly clocked SB's vs clocked other chips in games, and reviews.

Also... SB's didn't ALL clock high. A lot of people were stuck at 4.5ish. Claiming that they nearly all went to 4.8Ghz is BS.

Let's not forget the magic of the solder as well. Everything after SB appeared to overclock worse due to intel changing from solder to thermal paste. SB was great because Intel did the job properly, rather than cheaping out.

Its not ******** its about Coolign You had ui had bad clocker SO ?? Just checked my Sandybridge was running 1.58 vcore 24/7 for stable Overclock :]
But I do run External radiator that can cool down over 1000wats of heat :D

Did few runs 163 scores are on 102.6mhz BCLK and memory at 3282mhz
0c5786a4_Untitled.png
 
Just out of Interest having spooted my Cinebench XML still has my old FX-9590 score in it :D, (Vishera) @ 4.7Ghz: 110 Points

According to AMD the final IPC difference between Zen and their last CPU (Excavator) is 52%, no one knows the IPC difference between Vishera and Excavator so its not possible to know the IPC difference between AMD's previous 'Mainstream CPU' being Vishera. It looks like its a lot.

But is someone has a 4Ghz 1500X or above and provide a single threaded Cinenench score we could probably work it out, i say 1500X because the 1400, 1300 and 1200 are a good 10% slower, something to do with less Cache.

hbdfh.png
 
Its not ******** its about Coolign You had ui had bad clocker SO ?? Just checked my Sandybridge was running 1.58 vcore 24/7 for stable Overclock :]
But I do run External radiator that can cool down over 1000wats of heat :D

Did few runs 163 scores are on 102.6mhz BCLK and memory at 3282mhz
0c5786a4_Untitled.png

That'll do.... :D

Its 65.5%, Ryzen has 65% higher IPC than Vishera

4Ghz vs 4.7Ghz = 17.5%
Score difference = 48%

Add them together.
 
Its not ******** its about Coolign You had ui had bad clocker SO ?? Just checked my Sandybridge was running 1.58 vcore 24/7 for stable Overclock :]
But I do run External radiator that can cool down over 1000wats of heat :D

Did few runs 163 scores are on 102.6mhz BCLK and memory at 3282mhz

and... you had a good clocker?

Point is, just because you got a good chip doesn't make that the 'norm'. I am more inclined to believe that there are more people that will try very basic overclocking and stop when it starts being unstable, than those who are prepared to place expensive coolers/watercooling etc.

I'm not trying to get in a slapping contest here zeed. When the 2500k was around initially, a lot of people struggled to get to 5Ghz, some were lucky and managed it, others stuck much lower than that. I happened to get 4.5Ghz and never really pushed it as I only had a medium range cooler, and not the best MB.
 
Last edited:
5GHZ Sandy's weren't the norm. I did have a chip that would do 5GHZ, but at the same time I also had a crap chip that wouldn't do any better than 4.2GHZ.

4.6GHZ would have ideally been the sweet spot.
 
Am I missing something or are we getting confused between instructions per clock and instructions per second?

Surely if were trying to measure ipc then we should be matching clocks and not running one at 4.5-5.0ghz and others at 3.8-4.0ghz.
 
Am I missing something or are we getting confused between instructions per clock and instructions per second?

Surely if were trying to measure ipc then we should be matching clocks and not running one at 4.5-5.0ghz and others at 3.8-4.0ghz.

lol.. yes. Yes they are.
 
5GHZ Sandy's weren't the norm. I did have a chip that would do 5GHZ, but at the same time I also had a crap chip that wouldn't do any better than 4.2GHZ.

4.6GHZ would have ideally been the sweet spot.
Well besides me two other mates had 5ghz on sandy mine was worst out of them. As they not needed more than 1.48vcore to hit 5ghz could be due to later revisions tho since we waited for Faildozer after phenoms....
Think it was 3rd steppig we had.


Anyhow for gaming id pick up 5820k over ryzen its just faster and got less trpubles.
 
Also... SB's didn't ALL clock high. A lot of people were stuck at 4.5ish. Claiming that they nearly all went to 4.8Ghz is BS.
You are 100% right. I can't find the scores for my original 2500K, but it would not go beyond 4.4GHz regardless of how many volts was put into it upto 1.5 volts.

I could do a suicide run with the one below at 5.4Ghz @ 1.5v for super pi, that's about it, not stable at 5Ghz with 1.5v even.

EDIT: P67 chipset motherboards overclocked sandybridge CPU's better than Z68 and Z77 with lower volts, I can only assume it was because P67 didn't support integrated graphics.

I wish I had kept my Gigabyte Z77-UD5H, it would let my 2500K clock to 4.8GHz with 0.2v less than my MSI Z77 MPower, it still wasn't able to do 5GHz though fully stable.

MSI Z77 MPower
2500K: Batch No: 3138A838

4.8GHz = 1.326v load (prime95), 1.32v load (IBT 2.54) 1.352v idle. (Manual CPU Voltage = 1.36v BIOS)

MSI GD65 Z68 (G3)
2500K: Batch No: 3138A838

4.9GHz = 1.408 load (prime95), 1.408v load (IBT 2.54) 1.424v idle. (Manual CPU Voltage = 1.43v BIOS)
4.9GHz = 1.400v load (prime95), 1.42v load (IBT 2.54), 1.01v to 1.06v idle. (Auto CPU Voltage)
4.8GHz = 1.336v load (prime95), 1.344v load (IBT 2.54) 1.352v idle. (Manual CPU Voltage = 1.36v BIOS)
4.7GHz = 1.296v load (prime95), 1.296v load (IBT 2.54) 1.312v idle. (Manual CPU Voltage = 1.32v BIOS)
4.6GHz = 1.384v load (prime95), 1.408v load (IBT 2.54) 1.01v to 1.06v idle. (Auto CPU Voltage)
4.6GHz = 1.280v load (prime95), 1.280v load (IBT 2.54) 1.296v idle. (Manual CPU Voltage = 1.30v BIOS)


2500K: Batch No: L104A429
MSI GD65 Z68 (G3)

4.6GHz = 1.336v load (prime95), 1.36v load (IBT 2.54), 1.00v to 1.05v idle. (Auto or Manual CPU Voltage) (Maximum Clock stable)
4.5GHz = 1.328v load (prime95, 1.352v load (IBT 2.54), 1.00v to 1.04v idle. (Auto or Manual CPU Voltage)
 
Last edited:
Back
Top Bottom