Guy upgraded iMac 5K i5-7600 / 575 to i7-7700K and 2.5 GB TB SSD (with teardown video)

[EugW]

Some guy bought an i5-7600 iMac, tore it apart, and added a 7700K, 64 GB RAM, and additional 2.0 TB SSD on top of his 512 GB SSD.

Interesting project, but note that he made some weird claims in the video for some reason. First of all, he didn't buy a base model iMac like he claimed. He bought a mid-tier i5-7600. Second he claims his 7700K is running 8% faster than Apple's but that's simply false. His Geekbench multi-core scores for example are exactly in line with what everyone else is reporting. Perhaps he's just using a crappy random Geekbench score or else a Geekbench average score to represent Apple's 7700K.

But otherwise, it's an interesting video because it proves it can be done, and he also shows some of the teardown procedure (although it isn't an iFixit-type teardown summary).

 

[vkd]

Why don't you address your questions to him directly, on his Youtube video? No chance of a resolution by just speaking to the wind, is there?

 

[EugW]

Why don't you address your questions to him directly, on his Youtube video? No chance of a resolution by just speaking to the wind, is there?

This thread is not address those questions. I already made comments to address those questions in the comments section of his YouTube video. The purpose of this thread is to show his teardown, and to show the upgrade can be done.

 

[czacha]

Here in Poland you save about $250 with his method not including RAM and by choosing 1TB SSD. So it would be stupid to loose waranty for that amount of $

 

[EugW]

Here in Poland you save about $250 with his method not including RAM and by choosing 1TB SSD. So it would be stupid to loose waranty for that amount of $

Yeah, the 7700K upgrade here in Canada (edu) is all of US$240 done by Apple. So that upgrade afterwards voiding the warranty makes no sense whatsoever. The main cost savings he had was going for an add-on 2 TB SATA SSD, and adding 64 GB of aftermarket RAM. For these though, you can go with Thunderbolt SSD and of course almost everyone does aftermarket RAM, with no impact on warranty.

Ironically though, I personally prefer the 7600 over the 7700K. I had the 7700K and returned it because of the fan noise. But as with all things, YMMV.

 

[czacha]

I've read your posts. Somehowe I'm not affraid of fan noise. I'd rather set them to start spin even earlier than standard temp/speed factor.

 

[EugW]

So, does Coffee Lake 6-core require a new chipset or not? I have heard conflicting rumours on this. If not a new chipset then theoretically in a few months somebody could install an i7-8700K 6-core to get a ginormous boost in multi-core speed. The pinout is exactly the same as the 2017 iMac CPUs.

The i7-8700K is supposed to be 3.7 GHz with 12 MB cache, 6-core with 12-threads. I'd guess it does Turbo at 3.9 GHz multicore. If so, that's 23.4 GHz worth of clockspeed, vs the 7700K which has 17.6 GHz (or 4.4 GHz Turbo x 4). That represents a 33% (1/3rd !) improvement in multi-core speed.

Furthemore, it is supposed to be a 95 Watt TDP chip, which means it would work in the current iMac chassis, albeit with the same (or possibly worse) fan noise as the 7700K.

 

[czacha]

6-core CPU would be nice for video editors.. on Lightroom 4x 4.2GHz will be faster than 6x3.7GHZ.

 

[Erdbeertorte]

So, does Coffee Lake 6-core require a new chipset or not?

The i9, i7-7740X (faster 7700K "emergency update" to keep up with AMD) and the other i7-7whateverX with more than four cores already need the new Xeon chipset, so it might be very possible. If I remember correctly there is even an i5...X.


http://ark.intel.com/products/series/123588/Intel-Core-X-series-Processors

 

[EugW]

The i9, i7-7740X (faster 7700K "emergency update" to keep up with AMD) and the other i7-7whateverX with more than four cores already need the new Xeon chipset, so it might be very possible. If I remember correctly there is even an i5...X.

Yeah, but X uses a different socket.

 

[mpe]

The difference in speed the guy is experiencing is almost certainly caused by different RAM modules rather than CPU.

I am running aftermarket 2133 MHz modules in my 2015 iMac and can see a similar difference in benchmarks scores.

 

[EugW]

The difference in speed the guy is experiencing is almost certainly caused by different RAM modules rather than CPU.

I am running aftermarket 2133 MHz modules in my 2015 iMac and can see a similar difference in benchmarks scores.

Except it's not different. I used Apple RAM mixed with the Crucial RAM with the same specs, and got a score about 100 points off of his score. 100 points is 0.5%, not the 8% he was reporting, and 0.5% is within the margin of error for Geekbench.

 

[Erdbeertorte]

Yeah, but X uses a different socket.

That's what I meant. They might ditch the "old" socket for Coffee Lake and just name some or all X to K again. But who knows...

 

[EugW]

That's what I meant. They might ditch the "old" socket for Coffee Lake and just name some or all X to K again. But who knows...

Coffee Lake is already known to be using the same 1151 socket. What we don't know if it will work with the same chipset.

 

[Erdbeertorte]

Coffee Lake is already know to be using the same 1151 socket. What we don't know if it will work with the same chipset.

Sorry I didn't know that. Is it really already announced officially?

I just researched a little bit. It seems if the CPU fits into the socket it should also work with the chipset, maybe with some limitations.

In only know for sure a friend exchanged his i5-6600K on a Skylake board with an i7-7700K without any issue. So if that worked, Coffee Lake might work too. We'll see...

 

[propower]

I applaud the guy who did this! BUT...

2TB SSD inside the box runs at SATA III speed. External TB3 (orTB2) case for SSD runs at - SATA III speed. 4 drive TB2 and TB3 case is like $350 - runs SATA III speed and doesn't void any warranties. USB3 $25 case runs at 2/3 SATA III speed. No way its worth it in my book to open the case for that - and NO way it is remotely equal to a 2TB internal PCIe drive.

 

[mpe]

Except it's not different. I used Apple RAM mixed with the Crucial RAM with the same specs, and got a score about 100 points off of his score. 100 points is 0.5%, not the 8% he was reporting, and 0.5% is within the margin of error for Geekbench.

Not sure, but I think g.skill Ripjaws feature auto-overclocking RAM feature. At least they advertise it. That could explain the difference.

 

[bcortens]

Except it's not different. I used Apple RAM mixed with the Crucial RAM with the same specs, and got a score about 100 points off of his score. 100 points is 0.5%, not the 8% he was reporting, and 0.5% is within the margin of error for Geekbench.

I'd be interested to know what the boost clock was set to throughout the test. That's where I suspect the difference is coming from, though I really wouldn't think that simply using different thermal paste would cause such a difference...

 

[EugW]

Not sure, but I think g.skill Ripjaws feature auto-overclocking RAM feature. At least they advertise it. That could explain the difference.

I'd be interested to know what the boost clock was set to throughout the test. That's where I suspect the difference is coming from, though I really wouldn't think that simply using different thermal paste would cause such a difference...

Arrrgghh! As mentioned, there is no difference. His score is typical for the iMac i7 7700K. I got 202xx. He got 203xx. That's within the margin of error, as it's only about 0.5% difference, and I was just using stock RAM speeds. Furthermore, others with iMacs have gotten a little bit higher than his score too. The reason he says he says his is 8% faster is because he was comparing a number that was below 19000, when we already know from a bazillion tests out there that the iMac i7 77000 can hit more than 20000.

He probably was just looking at a Geekbench average or something, but the average includes really bad scores (eg. including anomalous scores, and scores on machines with too many background apps running).

 

[mmomega]

such a bad 7700K GeekBench

 

[SaSaSushi]

such a bad 7700K GeekBench

Veeeeeeery interesting scores indeed. I'd love to know how they were achieved.

I've got the exact same spec'd iMac and while I'm getting a few hundred points over the Geekbench baseline in single core and a few thousand points higher in multicore, they're still nowhere near those scores.

RAM alone would not seem to account for such a discrepancy.