I too think Apple will focus a bit more on GPU & gaming performance since Steam is out. Also, that article on the homepage suggests Apple is actively working on gaming performance. This is something Apple has never done before..
Got a tip for us?
Let us know
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
iMac: ATI 4850 vs ATI 5850
- Thread starter Paulyboy
- Start date
- Sort by reaction score
ok, thanks! but without averaging, the 5850 is about 5% faster. the 5870 is only faster because of higher clocks.. sad i know how it consumes EVEN more power. 5850 is apples bet bet to hold of until the northern islands 6xxx series comes out which will bring significant improvements to the mac/ATI RADEON line. 5850 seems to be one of those preperation. AMD took the 5xxx family to put in new technoligies etc but decreased or incrementely upgraded the speed of the desktop products not the mobility, in fact some of them are slower which is kinda a flop for amd
Hellhammer
Moderator emeritus
ok, thanks! but without averaging, the 5850 is about 5% faster. the 5870 is only faster because of higher clocks.. sad i know how it consumes EVEN more power. 5850 is apples bet bet to hold of until the northern islands 6xxx series comes out which will bring significant improvements to the mac/ATI RADEON line. 5850 seems to be one of those preperation. AMD took the 5xxx family to put in new technoligies etc but decreased or incrementely upgraded the speed of the desktop products not the mobility, in fact some of them are slower which is kinda a flop for amd
Southern Islands will focus on power efficiency but it doesn't guarantee better mobile GPUs. SI will unlikely introduce new cores thus unless ATI releases mobile Cypress, we won't see big increases in performance. Due power efficiency, I would guess that ATI does so though as even NVidia was able to provide "mobile" Fermi even though Fermi is a lot hotter. Mobile Cypress would be pretty expensive though.
ATI 4850 consumes more power than 5870 thus 5870 is possible in terms of power consumption but Apple might find it confusing.
hundert
macrumors regular
If you really have done a lot of research, you would choose i5 over i7 🙂
However, I do not know the prices in your country on the refurbs, so the i7 might be a nice catch. And even if it is:
i7 is good for rendering (as example). If you render on iMac with i7, it is faster, but mac with i7 gets much much much hotter, which will cause damage to display or other parts. That happens on the old macs too... black stains appear on the screen.
iMac is really not for rendering. And for occasional CPU power, i5 is enough 🙂 i7 is only in plus, when you do rendering for more than half an hour or other tasks that depend on CPU completely for longer period of time (that's not browsing, or looking movies, that's rendering mostly).
As for update, I don't think it will happen until end of October.
Your money, do what you want with it.
Here is the Thread about people who have black spots on Screen:
That is because of the heat. Macs have crappiest cooling system you can imagine.
https://forums.macrumors.com/threads/861355/
However, I do not know the prices in your country on the refurbs, so the i7 might be a nice catch. And even if it is:
i7 is good for rendering (as example). If you render on iMac with i7, it is faster, but mac with i7 gets much much much hotter, which will cause damage to display or other parts. That happens on the old macs too... black stains appear on the screen.
iMac is really not for rendering. And for occasional CPU power, i5 is enough 🙂 i7 is only in plus, when you do rendering for more than half an hour or other tasks that depend on CPU completely for longer period of time (that's not browsing, or looking movies, that's rendering mostly).
As for update, I don't think it will happen until end of October.
Your money, do what you want with it.
Here is the Thread about people who have black spots on Screen:
That is because of the heat. Macs have crappiest cooling system you can imagine.
https://forums.macrumors.com/threads/861355/
Hellhammer
Moderator emeritus
As far as I can tell, i5 and i7 are pretty equal in terms of heat
http://www.guru3d.com/article/core-i5-750-core-i7-860-870-processor-review-test/9
hundert
macrumors regular
I am an overclocker I know everything about i5 and i7
i7 gets much hotter than i5.
The reason Anandtech shows different Power Consumption for i7 and i5 is because the chips they used had different VID. Newer revisions of CPUs have lower VID.
VID is a Voltage specified by Intel for the chip. The voltage gets improved with new steppings or revisions.
In other words, they have a crappy i5 750 and a good i7 860.
Anandtech also makes errors. Things like TIM (themal grease) can get applied in a wrong way, the cooler might be mounted unstraight. However, there is no word about temperature in the article.
Here is link http://www.anandtech.com/show/2839/8
I had 8x i7 860 with different Batches and I had one with 62° and one with 49°C both under load. I remounted the cooler more than 5 times and applied TIM in different ways and one chip was hotter than the other.
In the Link you gave: They probably had mainboard lie down on a table, not in a closed case, like every test person does. That's why temps are so close.
i7 860 really gets 8-15°C hotter, depending on batch.
i7 gets much hotter than i5.
The reason Anandtech shows different Power Consumption for i7 and i5 is because the chips they used had different VID. Newer revisions of CPUs have lower VID.
VID is a Voltage specified by Intel for the chip. The voltage gets improved with new steppings or revisions.
In other words, they have a crappy i5 750 and a good i7 860.
Anandtech also makes errors. Things like TIM (themal grease) can get applied in a wrong way, the cooler might be mounted unstraight. However, there is no word about temperature in the article.
Here is link http://www.anandtech.com/show/2839/8
I had 8x i7 860 with different Batches and I had one with 62° and one with 49°C both under load. I remounted the cooler more than 5 times and applied TIM in different ways and one chip was hotter than the other.
In the Link you gave: They probably had mainboard lie down on a table, not in a closed case, like every test person does. That's why temps are so close.
i7 860 really gets 8-15°C hotter, depending on batch.
Hellhammer
Moderator emeritus
It makes no sense! Only difference is 130MHz i.e. a bigger multiplier plus Hyper-Threading. The chip is the same. With your logic, the i5 can also be a lot hotter, depending on your luck.
IMO, the 5850 without gddr5 is an inferior card compared with 4850 with gddr3 due to rops, texture units and memory bottleneck, apple should just admit its mobiity i wouldnt mind 🙂 😀
TMRaven
macrumors 68020
Altho power consumption directly transfers into heat, I don't think the manner in which thermal paste is applied affects power consumption.
That said, the chips are the same (i5 lynnfield is just i7 lynnfield with 20x133 multiplier instead of 21x133 and hyperthreading turned off) intel did it for marketing reasons.
And yes, according to your logic, an i5 could be hotter on any day if the thermal paste was applied badly. If you look over temperature threads, you'll notice that people who post temps for their i5s and i7s are virtually the same.
That said, the chips are the same (i5 lynnfield is just i7 lynnfield with 20x133 multiplier instead of 21x133 and hyperthreading turned off) intel did it for marketing reasons.
And yes, according to your logic, an i5 could be hotter on any day if the thermal paste was applied badly. If you look over temperature threads, you'll notice that people who post temps for their i5s and i7s are virtually the same.
hundert
macrumors regular
It is hyper threading alone that is reason for the heat. On Windows you can turn HT off in bios and chip gets much cooler.
Yes, i5 can be hotter as I stated. But in average the i7 will be 10° kelvin hotter.
I wish I could prove it to you, but I don't want to void the warranty putting the i7 in instead of i5. There is a sticker that prevents me from doing it.
lpolarityl
macrumors 6502a
Some very mixed responses in this thread. As an actual hardware reviewer, let me add my two cents on what the differences are between the i5 and i7. Most important is Hyper Threading. i5 750 (4 cores/4 threads) has none, while the i7 860 (4 cores/8 threads) does. Automatically, the i7 860 will consume more power and in turn produce more heat. This is a side effect of Hyper Threading, but more threads tend to equal more heat with Intel's architecture.
The only reason to spend the extra cash and get the i7 setup over the i5 is simple, demand. If you have a high demand for multi-threaded jobs (encoding, very heavy multitasking) then the i7 option is for you. If you don't have a high need for multiple threads, then go with he i5. There aren't many apps that will utitilize all 8 threads but the ones that do will certainly show performance improvements.
The 4850 is an older GPU and there is no reason to get that over a 5850 no matter what platform you're using (PC, Mac, mobility options, etc.). If Apple is hinting towards a 5850 GPU powered iMac, and you have the patience to wait it out, then by all means wait. The 5850 is a newer generation altogether supporting new features like DX11 (BootCamp), uses less power and overall has higher performance. If the 5800 series is just now coming to Apple products, the chance of 68xx series GPU's coming anytime soon is unlikely. Considering the 4800 series is still being offered on now iMac's and Mac Pro's is a sign of slow progression so a 6800 series by years end is unrealistic.
I've built many PC's and overclock the living crap out of them. I got my AMD six core 1090T to 4.5GHz from 3.2GHz. I got my i5 750 2.66GHz to 4.4GHz, and my i7 920 (D0 Stepping) 2.66GHz to 4.5Ghz. So overclocking wise, i5's and i7's are fantastic, but there is no doubt that the Hyper Threaded capable CPU's use more power and produce more heat. Throughout my reviews and overclocking endeavors, I noticed about a 10C difference between the i5 750 and i7 920 when under load. AMD however, are icebergs in comparison, even the six core at load runs 20C+ cooler than an i5 or i7 at similar clocks.
It's too bad Apple didn't partner with AMD, imagine a 6 core iMac with a 5850 for encoding videos and gaming. It would be cheaper, it would run very cool, and it would produce nice encoding results because more physical cores is always better when encoding. I know in the PC world the 5850 makes the 4850 seem like a piece of crap, especially when you factor in the DX11 features and AA/AF improvements. And of course, the monthly ATi driver releases that almost always improve performance.
The only reason to spend the extra cash and get the i7 setup over the i5 is simple, demand. If you have a high demand for multi-threaded jobs (encoding, very heavy multitasking) then the i7 option is for you. If you don't have a high need for multiple threads, then go with he i5. There aren't many apps that will utitilize all 8 threads but the ones that do will certainly show performance improvements.
The 4850 is an older GPU and there is no reason to get that over a 5850 no matter what platform you're using (PC, Mac, mobility options, etc.). If Apple is hinting towards a 5850 GPU powered iMac, and you have the patience to wait it out, then by all means wait. The 5850 is a newer generation altogether supporting new features like DX11 (BootCamp), uses less power and overall has higher performance. If the 5800 series is just now coming to Apple products, the chance of 68xx series GPU's coming anytime soon is unlikely. Considering the 4800 series is still being offered on now iMac's and Mac Pro's is a sign of slow progression so a 6800 series by years end is unrealistic.
I've built many PC's and overclock the living crap out of them. I got my AMD six core 1090T to 4.5GHz from 3.2GHz. I got my i5 750 2.66GHz to 4.4GHz, and my i7 920 (D0 Stepping) 2.66GHz to 4.5Ghz. So overclocking wise, i5's and i7's are fantastic, but there is no doubt that the Hyper Threaded capable CPU's use more power and produce more heat. Throughout my reviews and overclocking endeavors, I noticed about a 10C difference between the i5 750 and i7 920 when under load. AMD however, are icebergs in comparison, even the six core at load runs 20C+ cooler than an i5 or i7 at similar clocks.
It's too bad Apple didn't partner with AMD, imagine a 6 core iMac with a 5850 for encoding videos and gaming. It would be cheaper, it would run very cool, and it would produce nice encoding results because more physical cores is always better when encoding. I know in the PC world the 5850 makes the 4850 seem like a piece of crap, especially when you factor in the DX11 features and AA/AF improvements. And of course, the monthly ATi driver releases that almost always improve performance.
lpolarityl
macrumors 6502a
Its gonna be hotter iMac or not. The i7's have more threads so its going to consume more power. It's basically what I explained above. The iMac uses a 1156 socket i7 860, not a 1366 socket i7 920. The Mac Pro towers use the i7 920's which uses triple channel DDR3 memory controller while the 1156 socket (i5/i7 8xx) uses a dual channel DDR3 memory controller.
Are you trying to tell people NOT to get the i7 because its hotter and will therefore, wear out their Mac quicker?
Hellhammer
Moderator emeritus
Many AMD 1090T have wrongly calibrated heat sensor... For example, Phenom x4 965 have ~15c higher temps than x6 even though they are based on same architecture and both are 45nm. In power consumption, 965 uses several watts less too. It's very likely that the temps you have got aren't right. Lynnfields consume even less power than Phenom x4s.
Also, at TMraven said, you cannot compare 130W CPU to 95W
TMRaven
macrumors 68020
Yes, we all know this. I'd like to see more test numbers that actually prove the i7 860s in the iMacs to consume more power and thus run hotter than the i5 750s though. Hellhammer's graph from 3d guru shows the two to consume the same amount of power when encoding in x264-- which is hyperthreaded. Hundert said it might be flawed due to differences in thermal paste, but does the application of thermal paste really affect power consumption? (we know it could affect heat, but does it affect power consumption)
lpolarityl
macrumors 6502a
No, not at all. I'm just saying, if you're going to buy an iMac, pick the one that suits your needs instead of thinking the "top dog" choice is the one to purchase. I don't know the current situations with 27" iMac running i7's, but if factual proof is coming out stating i7 powered iMac's are producing heat issues and screen defects, I'd certainly recommend staying away and getting the cooler running i5 offering.
Wrongly applying thermal paste won't increase power consumption, if anything it would decrease it because the CPU would throttle down its speed to keep itself from getting too hot. Both AMD and Intel have that killsafe feature would usually shuts the machine down if it gets TOO hot. So no, it won't increase power usage but it will definitely increase temperatures because of uneven contact with the heatsink.
Depending on what app was used, x264 encoding in that particular benchmark/app might have not been utilizing all 8 threads. This is where things get a little harry. When I do my reviews and when most sites do their reviews, each site has their own flavor of benchmarks. I use linpack based programs to stress the processor as much as possible. Either it be OCCT, LinX or Intel Burn Test, all do a great job of using calculations to completely peg every thread the CPU has. When I've personally done my reviews, a i7 920 with Hyper Threading enabled was about 8-10C hotter than when running the same CPU with Hyper Threading disabled. In comparison to an i5 750 at same clocks though, its at least a 10C difference.
I know all this already, I've personally reviewed all those CPU's. I know about the temperature readings being off but I also use an infrared thermometer so I'm also still right with what I said about my temperature experiences.
Wattage has nothing to do with performance scaling, just compare 140W Phenom's to 130W i7's, clock for clock Intel's do more instructions per cycle, but thats all due to architecture. Still, AMD's perform well and are much cheaper, too bad Apple didn't dip into AMD products cause we'd all be paying less for Apple products if they did.
Hellhammer
Moderator emeritus
Temperature goes hand in hand with power consumption. AMD may have better cooler or it may spin faster then but the x6 isn't cooler than Lynnfields
I never said about performance... iMac uses i7 860 which is 95W thus you cannot compare it to the 920 and say the i7 is at least 10c hotter.
lpolarityl
macrumors 6502a
X6's (Thuban) run much cooler than Lynnfields. I have a water cooled setup right now where I switched from my i5 750 @ 4.0 to my X6 1090T at 4.0+ (currently 4.2) and it runs 15C cooler on average at full load. Again, personal experience.
You can compare a i7 860 to an i5 750 and you will still see a temperature difference, about 8-10C. I've seen it first hand. With air cooling especially, there is about a 8-10C difference between the i5 750 and i7 860 when fully loaded. When I say utilized, I'm referring to all 8 threads being used on the 860 and all four being used on the 750. Not all benchmarks can utilize 2+ threads (100%), yet alone 8. Encoding is why people by the i5 and i7 powered iMacs. Depending on the app, it will or will not use all available threads. When it does use ALL threads, you will see a 8-10C difference between the 860 and 750, again, this is just my personal experience.
Hellhammer
Moderator emeritus
Dare to share some evidence then? If you review them, you must have a site with nice graphs to show? How about fan speeds then? Were they set to certain level? i7 should also have a blasting fan too if it runs hotter. Also, it doesn't run hotter if all threads aren't in use. I don't mind if it's little hotter if it provides 30% better performance when I need it. Idle temps should be the same because only one or two threads are in use
Have you tested Phenom 965? Its temps should be comparable to 1090T (same power consumption). No chance 1090T runs cooler than x4 965 under full load
lpolarityl
macrumors 6502a
I have plenty of proof, I review remember? 😎
BTW, I'm not posting just to annoy you, I'm posting because everything you're saying is confusing people, your not stating facts, just hear say, in my opinion.
Thuban (X6) at 4.0GHz (water cooled)
i5 750 at 4.0GHz (water cooled)
i7 920 at 4.2Ghz (best air cooler available (Noctua NH-D14), but it was 65F in the room at the time, temps were very generous)
And here is a link to reviews I've done:
http://bjorn3d.com/content.php?author=John+Armstrong&cat=9999&keyword=&Submit=Go&viewcat=1
So, look at the pics and look at the maximum temperatures. x6 in the 30-40's, i5 in the mid to high 60's, and i7 in the mid to high 70's. Even though the Phenom heat sensors are off 10c (thats the current agreement from what ive seen), then lets say my x6 is really in the 40's to 50's. Even if that were 100% true, it still runs 10-15c cooler than an i5 750. You mentioned X6's didn't run cooler than Lynnfields (i5/i7 8xx), they do run cooler. As for Phenom 965's, I wouldn't know I've never had a 965. I had a Phenom II 920 but that was long ago and won't really compare to these newer steppings/revisions.
Now, the i7 is in the 70's, same exact clock speed as the i5 750, and its still over 10C hotter. Now, I know the i5 was water cooled, but my water cooling loop cools the MOSFET's/VRM's, 5850 GPU, SouthBridge, and the CPU. So really, it only about 2-3c better than the i7 on air. There's your proof. 🙂
I forgot to mention that once overclocking is put into the picture, wattage ratings mean diddly squat. Once you push the CPU passed its native clock speed, voltages are increased and the TDP rating becomes null. This is why I know the i5's run an average of 8-10C cooler at similar clock speeds. This also relates down to default clocks and ratings, once a CPU is pegged it wills how its true temperatures.
lpolarityl
macrumors 6502a
I don't have power consumption numbers (on hand) unfortunately, just temperature results. I'll have to see if I can find any for ya.
Side note: We really should get this back to the original topic, we pissed all over the OP's question, sorry OP!
To the OP: Wait for a 5800 series powered iMac if its a possible future option!