Broadwell i7 5775C Gaming Performance on par with Haswell

[flibberdipper]

I prefer the "k"s over the "c"s, but whatever.

I have a Core i7 5770C. What does the "C" stand for? Clocks.

[duriel]

i would fucking hope so?

 

but its inability to pass 4.3ghz at 1.4v is atrocious.

i'm sitting at 4.5ghz 1.170v right now.. i could *easily* push my chip to 5ghz if i was comfortable with pushing 1.4v on my chip which would match/beat the 5775c.

 

no reason to buy broadwell chips unless you're running a SFF computer with no discrete GPU.

 

 

I have a Core i7 5770C. What does the "C" stand for? Clocks.

 

 

C stands for Crystalwell

[patrickjp93]

You still don't have a SSD? *gasp*

You can buy a 60GB one so cheap for a long time as a boot drive.

Due to the fact I need to dual-boot and there are a number of programs out there which you can't move to a secondary drive, I need something a fair bit larger. Luckily my 3-drive RAID 0 is clipping along nicely, but I know how much better it could be.

[patrickjp93]

i would fucking hope so?

but its inability to pass 4.3ghz at 1.4v is atrocious.

i'm sitting at 4.5ghz 1.170v right now.. i could *easily* push my chip to 5ghz if i was comfortable with pushing 1.4v on my chip which would match/beat the 5775c.

no reason to buy broadwell chips unless you're running a SFF computer with no discrete GPU.

C stands for Crystalwell

I think we should reserve overclocking judgment for now. You can't even buy them yet. The first couple that have had benches revealed could very well be duds in the silicon lottery.

[MageTank]

I think we should reserve overclocking judgment for now. You can't even buy them yet. The first couple that have had benches revealed could very well be duds in the silicon lottery.

Not only that, but the BIOS revisions also lack the proper microcode for these CPU's. I got my G3258 very early, and needless to say, my OCing improved drastically on my AsRock Z97m-ITX/AC after later BIOS revisions started coming out. The board matters less these days when OCing, but they still provide enough of an impact to determine how much voltage one might need for something to be stable. 

 

 

i would fucking hope so?

 

but its inability to pass 4.3ghz at 1.4v is atrocious.

i'm sitting at 4.5ghz 1.170v right now.. i could *easily* push my chip to 5ghz if i was comfortable with pushing 1.4v on my chip which would match/beat the 5775c.

 

no reason to buy broadwell chips unless you're running a SFF computer with no discrete GPU.

 

 

 

 

C stands for Crystalwell

 

You are comparing what i consider to be a very solid bin, to something that could easily be a terrible bin. We do not have any other broadwell chips to compare this OC to, as the others that ran iGPU benches did not OC the CPU itself. 4.5ghz on 1.17V is very impressive considering i need 1.17V to hit 4.3ghz on my G3258, and my G3258's stock VID is 1.080. 

 

As for people worried about building a new PC, and choosing broadwell over skylake, i would say if you are not in a hurry, or that you don't exactly NEED an upgrade this early, to wait to see what Skylake has to offer.

 

So far, we only have rumors, but Intel themselves have specifically said that "Skylake is similar to going from Prescott to Conroe". Conroe had like, 3x the IPC of Prescott if i remember correctly. While i never take claims seriously when it comes from the companies themselves, there are some things that help lean towards their claims. We know that AVX3 instruction sets were done away with on the normal consumer Skylake chips, but float units were still kept at 512-bit, without the instruction sets, but maintaining the AVX3 registers, they would be able to do 32 AVX2 things per clock. That is a ton of math power if one actually thinks about it, compared to what we currently have available to us. 

 

Think about it. AVX3 (ZMM) is twice as large as AVX2 (YMM). 512-bit FP registers would be huge, lol. This very same theory is what makes me excited about Zen. Zen is still using the dual FMAC float unit design that was originally used in Bulldozer, but doubled the size of the FMAC bit-width. That float unit was designed to be a shared resource, but with Zen, it won't be shared between Int units. That would yield large efficiency increases.

 

Sorry for the long and semi-off topic tangent, just wanted to share the thoughts i was having on other processors coming out soon. Either way, the future for all of the CPU's being released as of now is looking pretty bright. As long as you build around what they were designed for, you really cannot go wrong.

[patrickjp93]

Not only that, but the BIOS revisions also lack the proper microcode for these CPU's. I got my G3258 very early, and needless to say, my OCing improved drastically on my AsRock Z97m-ITX/AC after later BIOS revisions started coming out. The board matters less these days when OCing, but they still provide enough of an impact to determine how much voltage one might need for something to be stable. 

 

 

 

You are comparing what i consider to be a very solid bin, to something that could easily be a terrible bin. We do not have any other broadwell chips to compare this OC to, as the others that ran iGPU benches did not OC the CPU itself. 4.5ghz on 1.17V is very impressive considering i need 1.17V to hit 4.3ghz on my G3258, and my G3258's stock VID is 1.080. 

 

As for people worried about building a new PC, and choosing broadwell over skylake, i would say if you are not in a hurry, or that you don't exactly NEED an upgrade this early, to wait to see what Skylake has to offer.

 

So far, we only have rumors, but Intel themselves have specifically said that "Skylake is similar to going from Prescott to Conroe". Conroe had like, 3x the IPC of Prescott if i remember correctly. While i never take claims seriously when it comes from the companies themselves, there are some things that help lean towards their claims. We know that AVX3 instruction sets were done away with on the normal consumer Skylake chips, but float units were still kept at 512-bit, without the instruction sets, but maintaining the AVX3 registers, they would be able to do 32 AVX2 things per clock. That is a ton of math power if one actually thinks about it, compared to what we currently have available to us. 

 

Think about it. AVX3 (ZMM) is twice as large as AVX2 (YMM). 512-bit FP registers would be huge, lol. This very same theory is what makes me excited about Zen. Zen is still using the dual FMAC float unit design that was originally used in Bulldozer, but doubled the size of the FMAC bit-width. That float unit was designed to be a shared resource, but with Zen, it won't be shared between Int units. That would yield large efficiency increases.

 

Sorry for the long and semi-off topic tangent, just wanted to share the thoughts i was having on other processors coming out soon. Either way, the future for all of the CPU's being released as of now is looking pretty bright. As long as you build around what they were designed for, you really cannot go wrong.

AVX isn't just floats for the record. 

[Techhog]

I can't wait until Linus & Luke review this, hopefully by then the OEMs will have released proper BIOSes so that these issues that Maximum PC encountered don't happen.

I'd really like to see how good it will overclock, with the iGPU disabled and that 128 MB of eDRAM dedicated entirely to the CPU as a L4 cache. From what I've read from other reviews the average ceiling seems to be 4.2 GHz.

4.2GHz is my conservative OC for my 4670k...

[MageTank]

AVX isn't just floats for the record. 

Won't stop me from being super excited nonetheless.

[patrickjp93]

Won't stop me from being super excited nonetheless.

Isn't it nice that we can have a civilized conversation?

[MageTank]

Isn't it nice that we can have a civilized conversation?

https://en.wikipedia.org/wiki/Christmas_truce

 

Not exactly uncommon, but yes, its always nice.

[RagnarokDel]

I have haswell I5 4670k and the only reason I am considering upgrading to Skylake is to get a I7.

[Lays]

God I hope it doesn't take those sorts of voltages to reach clock speeds like that.

[bcredeur97]

Only real difference is power consumption I guess... But I'm ok with that because it's still very strong

[ImBleu]

 

What does the name matter if it's still got an unlocked multiplier?

 

Familiarity for me Nothing else.

[MageTank]

New Update: We now have something to compare MaximumPC's overclock to. 

 

http://www.legitreviews.com/intel-core-i7-5775c-broadwell-processor-review_166875

 

Page 3:

 

 

We overclocked the Intel Core i7-5775C processor by simply increasing the multiplier and raising the voltage on the processor when needed. The UEFI microcode for enabling eDRAM selections and overclocking was not available in the build for the board that we have, so we left that alone for the time being (We’ll do an article on that later). By just raising the multiplier we were able to get to 4.1 GHz without needing to increase the voltage. To hit 4.2 GHz we had to increase the voltage to 1.3V and then to hit 4.3GHz we had to increase it again to 1.35V. We were able to get 4.4 GHz up and running on the desktop, but it would crash in very CPU intensive benchmarks. We went up to 1.425V on the processor and couldn’t get it stable at 4.4 GHz before calling it quits and we just benchmarked everything in our test suite at 4.3GHz.

We later learned that you can put more voltage to the processor without blowing it up right away and was able to get 4.4GHz stable at 1.475V. If you want to run 4.6-4.7GHz on one of these processors you need to run around 1.6V on the core and that is something we weren’t interested in doing and you’ll see why when we get to the power consumption section!

 

Another interesting thing to note: 

 

 

 

Benchmark Results: The Intel Core i7-5775C processor is a 65W TDP processor and at idle the entire system was pulling 26.2 Watts and the wall and when under fill load we were pulling 79.0 Watts. All that efficiency is thrown out the window with the overclock though and our idle power increased to 34.6W and at load we were now hitting 119 Watts, which is similar to a Intel Core i7-4770K Quad-Core processor with default settings. We used Handbrake to rip a full 1080P Blu-Ray as our load test, so this is representative of a real world scenario. So, to get a 50% overclock and performance increase we needed to more than double the power consumption at load.

 

 

 

When it comes to overclocking we were easily able to take the Intel Core i7-5775C up to 4.3GHz by simply increasing the CPU multiplier up to 43 and bumping the voltage up to 1.35V. The overclock was rock solid and out CPU temperature never broke 70C, but the negative was that it threw the power efficiency all to heck.  Our 16.2% overclock from the Turbo clock rating had us using 54.4% more power at load. It also used more at idle and since we were using more power it was running hotter and that increases the fan noise as everything ramps up to compensate. So, overclocking is easy on the Core i7-5775C, but you basically take your 65W TDP processor and turn it into one that is likely pushing 80W or more depending on the clock speeds and voltage you end up using. That sounds bad, but you basically just end up where the high-end Haswell parts are at.

[Agosto]

Strangely, it has a VERY high stock voltage for a 14nm chip. And 1.4V for only 4.4 GHz is too much for such a small node...

 

I hope Skylake will be better. These are clearly low binned chips, while better ones go to laptops

[MageTank]

Strangely, it has a VERY high stock voltage for a 14nm chip. And 1.4V for only 4.4 GHz is too much for such a small node...

 

I hope Skylake will be better. These are clearly low binned chips, while better ones go to laptops

Yeah, I think Broadwell is just a test subject for this process. Though, if you take into consideration the stock voltage, and the clock it can achieve without touching voltage at all, it's still not too bad for what it is. It might still be too early to judge broadwell, as we only have 2 samples that were OC'd by separate sources, but if these are "good" bins, then i would be inclined to worry.

 

When i get my hands on a Broadwell chip, i will probably be going the opposite direction anyways. An undervolt for SFF to see how low i can go while still remaining stable on stock speeds.

[Agosto]

Yeah, I think Broadwell is just a test subject for this process. Though, if you take into consideration the stock voltage, and the clock it can achieve without touching voltage at all, it's still not too bad for what it is. It might still be too early to judge broadwell, as we only have 2 samples that were OC'd by separate sources, but if these are "good" bins, then i would be inclined to worry.

 

When i get my hands on a Broadwell chip, i will probably be going the opposite direction anyways. An undervolt for SFF to see how low i can go while still remaining stable on stock speeds.

1.225 V at stock is a lot for such a small chip, considering it's on 14nm FinFET. The 4790K has less than 1.2V at stock... and it runs at 4/4.4 GHz instead of 3.3/3.7 GHz.

VERY low binned chips with high stock voltage, to be sure they don't get unstable.

I'm pretty sure you will find out that you can't undervolt very well - at least it has very limited power consumption even at that voltage.

[IceCold007]

Like ive said on other websites. A general rule when upgrading CPU for gaming benefits only. Upgrade least 2 or 3 generations apart. My i7 4770k will be good enough until skylake or the gen after that. Then again if the rumours are true about skylake the gaming performance for that CPU might be worth a upgrade only time will tell but if its only 3% increase from Broadwell well I can wait

[BobbyG2]

When are these coming out? Debating on getting one of these i7s or just going x99 for a matx build next month.

[lightningterror]

I have a Core i7 5770C. What does the "C" stand for? Clocks.

 

Desktop Processor based on the LGA 1150 package with high performance graphics

 

 

source

http://www.intel.com/content/www/us/en/processors/processor-numbers.html

[flibberdipper]

source

http://www.intel.com/content/www/us/en/processors/processor-numbers.html

Mine is better.

[johnleyo]

thank you 

[Ketcchup]

The hell are you expecting, it's a Tock, not a Tick so there's only die shrink

[MageTank]

The hell are you expecting, it's a Tock, not a Tick so there's only die shrink

Who exactly are you talking to? Everyone knows its a Tock, but it was assumed by a large amount of people that Broadwell would perform worse than Haswell due to the much lower clock speeds. This thread is only showing that Broadwell is performing on par with Haswell, which is putting those original doubts to rest. You are also incorrect, it's not just a die shrink. A lot has been changed with Broadwell. 

 

The EU's in Broadwell more than doubled what Haswell had, from 20 to 48, making the Iris Pro 6200 far more superior than anything Haswell had to offer. Broadwell also got a huge chunk of eDRAM (128mb) that acts as L4 cache. It also has slightly less L3 cache than Haswell. The only thing that came as expected from a die shrink was the lower TDP. The rest was entirely different.

 

There is no doubt that upgrading from a current Haswell i5 or i7 to a Broadwell i5 or i7 would make no sense, and that one should just stick with what they have if they are planning to do such an upgrade. However, someone that has a G3258 or i3 could still find a great upgrade in Broadwell. Regardless, Haswell will always have the clock advantage over Broadwell, but thanks to the slight IPC increase, even if Haswell has a clock advantage, Broadwell will not be too far behind in performance, but have extra features that will make it just as appealing for new buyers.

 

If only Intel would have released Broadwell sooner, then it wouldn't be facing such a little lifespan in the market.