Intel Haswell 35W Dual-Core i5-4570T Benchmarks Surface Online

[Over_Lord]

While Intel's quad-core Haswell architecture based 4th generation Core processors are already out there dominating benchmark charts (even if only by a small margin) and scoring new accolades in power efficiency, the dual-core Core processors have been left in the dark. At least, in the desktop segment. Not anymore though, as PC Games Hardware managed to score a sample of the energy efficient Core i5-4570T, a dual-core Haswell based processor for the desktop with a very impressive, and low, 35W TDP only.

Here's a quick look at the Haswell desktop processors line-up:



While the Haswell desktop SKU in question is an energy efficient dual-core part, it doesn't fall short of delivering satisfactory performance. It manages to edge competing solutions from AMD, with more number of cores and many factors higher TDP, in some of the benchies. Check out a couple of them, posted below for your viewing pleasure.



For the complete benchmark results, hit the source link posted below.

View at TechPowerUp Main Site

 

[Sempron Guy]

This will suit best on a gaming htpc rig. It would've been nice though if they'll create a quad version on the same tdp.

 

[Fourstaff]

This will suit best on a gaming htpc rig. It would've been nice though if they'll create a quad version on the same tdp.

The clockspeed will need to be severely gimped to fit a quadcore at 35w. I would prefer a dualcore with faster clockspeed and 2c/4t

 

[Sempron Guy]

Based on reviews Haswell seems to scale well even on lower clocks? Strong single thread performance but it suffers when the game demands more than two cores based on the review.

 

[dj-electric]

without the iGPU you have a good CPU under about 26W, can't really ask for more.

 

[Fourstaff]

Based on reviews Haswell seems to scale well even on lower clocks? Strong single thread performance but it suffers when the game demands more than two cores based on the review.

I haven't seen the scaling below 3Ghz on Haswell, Ill take your word for it. I wonder how far the clock needs to go before reaching the required TDP

 

[Ikaruga]

Nice. Wish they would do a desktop CPU like this with a HD5000 as well.

 

[TheLaughingMan]

I love how the AMD FX processors are limited to a 970 board while Intel gets an entire array of high-end to mid-ranged boards. Even more amazing how this lowers the Hyper Transport Speed by 200 MHz (400 MT/s). And then they are tested on a 2 year old game with badly optimized DX11 code.

Yeah, these results are totally unbiased and relevant to current gaming use for these chips.

 

[TheinsanegamerN]

I love how the AMD FX processors are limited to a 970 board while Intel gets an entire array of high-end to mid-ranged boards. Even more amazing how this lowers the Hyper Transport Speed by 200 MHz (400 MT/s). And then they are tested on a 2 year old game with badly optimized DX11 code.

Yeah, these results are totally unbiased and relevant to current gaming use for these chips.

oh, you mean AMD only performs well under specific settings, using only a certain chip-set and board, with only games optimized for them, while the intel chips perform well with a VARIETY of boards and games?

outside of battlefield 3, there are maybe 1 or 2 games that run better on amd than intel. the motherboard doesn't have much of an effect on that. fx chips are just slower. welcome to reality.

 

[Patriot]

oh, you mean AMD only performs well under specific settings, using only a certain chip-set and board, with only games optimized for them, while the intel chips perform well with a VARIETY of boards and games?

outside of battlefield 3, there are maybe 1 or 2 games that run better on amd than intel. the motherboard doesn't have much of an effect on that. fx chips are just slower. welcome to reality.

Just because they are slower doesn't mean they should be handicapped. However this is more a matter of target market than anything else.
Intel limits the chipsets you can use if you want to overclock so by default they get better boards. Everyone knows AMD is going to lose so no one bothers about getting a high end board especially for a mid or low end chip. The top end chips tend to get nice boards so meh.

I really don't see someone buying a 990fx board for an amd FX4xxx.

 

[Roph]

Shame you're forced to also buy/run a bad GPU if you get one. Take out the GPU, add 2 more clores, 45W chip maybe?

 

[Aquinus]

Shame you're forced to also buy/run a bad GPU if you get one. Take out the GPU, add 2 more clores, 45W chip maybe?

Except the flaw with your logic is that I'm pretty sure that Intel power gates the iGPU when it's not being used.

 

[TheLaughingMan]

Just because they are slower doesn't mean they should be handicapped. However this is more a matter of target market than anything else.
Intel limits the chipsets you can use if you want to overclock so by default they get better boards. Everyone knows AMD is going to lose so no one bothers about getting a high end board especially for a mid or low end chip. The top end chips tend to get nice boards so meh.

I really don't see someone buying a 990fx board for an amd FX4xxx.

I don't see anyone buying an Z77 or Z87 board for a i5-4570T, but didn't stop them from testing with one of those. This is benchmarking. Its not about what people may or may not do. Its about comparing two products with as few variables as possible under the best light.

And plenty of people do exact that. the diff. in price for a 990FX board and a decent 970 is only about $25. Future proofing as well as planning for future upgrade.

oh, you mean AMD only performs well under specific settings, using only a certain chip-set and board, with only games optimized for them, while the intel chips perform well with a VARIETY of boards and games?

outside of battlefield 3, there are maybe 1 or 2 games that run better on amd than intel. the motherboard doesn't have much of an effect on that. fx chips are just slower. welcome to reality.

No. My point was they are comparing a chip being tested on a top-end board even though it is at default settings. While the AMD counterparts are being tested, as stated above with a handicap. High-end boards from Intel limit overclocking options and differ in features. AMD boards actually differ in performance a 400 MT/s performance difference is huge since that applies too all board communications between the processor and the RAM and Northbridge.

And in reality, when comparing modern games the diff. between AMD and Intel is often minor or non-existent. FX chips are slower overall, but Vishera has proven to hold its in multi-threaded tests.

In the end, I am not trying to say FX chips are better than Intel. I am saying these results are bias and should not be given much weight.

Real Interesting when I change the test game on that site the results tell a different story:

 

[newtekie1]

So why are they calling this an i5-4570? This is an i3, and it shouldn't have the same number as a Quad-Core part, it will just confuse the customer.

 

[Fourstaff]

So why are they calling this an i5-4570? This is an i3, and it shouldn't have the same number as a Quad-Core part, it will just confuse the customer.

Intel names i3/i5/i7 based on performance for a given TDP, not core count. They have been doing this for years now.

 

[newtekie1]

Intel names i3/i5/i7 based on performance for a given TDP, not core count. They have been doing this for years now.

That doesn't make it right.

 

[Fourstaff]

That doesn't make it right.

It makes sense though, the naming system is consistent from 100+w chips all the way down to the 17w low power ones.

 

[ViperXTR]

intel made some dual core i5's in the desktop in its 1st gen yes? (i5 6xx ones)

 

[newtekie1]

It makes sense though, the naming system is consistent from 100+w chips all the way down to the 17w low power ones.

Not really, if it is going to be called an i5, don't give it the same model number as a quad-core part, give it a lower model number.

intel made some dual core i5's in the desktop in its 1st gen yes? (i5 6xx ones)

Yes, and it made sense then because the i5-7XX was quad-core and i5-600 was dual-core w/ HT. Now, IMO, i5 should be quad-core always and i3 should be dual, at least in the desktop market.

 

[Fourstaff]

Not really, if it is going to be called an i5, don't give it the same model number as a quad-core part, give it a lower model number.

Yes, and it made sense then because the i5-7XX was quad-core and i5-600 was dual-core w/ HT. Now, IMO, i5 should be quad-core always and i3 should be dual, at least in the desktop market.

Well this part is more closely related to laptop parts than desktop, personally I think the number system makes sense. It might not make sense to you since you don't really care much about mobile parts (or anyone who doesn't keep up up to date with mobile).

 

[newtekie1]

Well this part is more closely related to laptop parts than desktop, personally I think the number system makes sense. It might not make sense to you since you don't really care much about mobile parts (or anyone who doesn't keep up up to date with mobile).

Closely related to mobile part? Why because it uses low power? No. It is a desktop part, so it should fall in line with desktop naming.

 

[cdawall]

Doesns't seem like a bad chip for the TDP it pulls. Why they compared the AMD in a low end board I don't know nor do I care since its the norm.

 

[Stickmansam]

i5 vs i3

i5 are called i5's instead of i3's even if they have the same core count is a strategy from mobile use. In mobile, i5's are only differentiated mainly by having turbo boost enabled. I guess they are slowly bringing that naming convention over to desktop. In mobile there are 2c/4t i7's. Not saying that this is right but this is what I think they are basing their decisions on. Sort of like the i5 7xx vs i5 6xx. The difference in i5 was not only the change in core count but also the process was different. This is going to be a problem since intel judges their chips differently desktop vs mobile based on realtive performance of the platform to name the chips rather than based on core-count.

 

[Prima.Vera]

That doesn't make it right.

Agree. I almost got pranked by intel on the laptop market because I didn't know that ALL the i5 mobile CPUs are in fact dual-cores. Crappy intel!!! :shadedshu

 

[Aquinus]

Agree. I almost got pranked by intel on the laptop market because I didn't know that ALL the i5 mobile CPUs are in fact dual-cores. Crappy intel!!! :shadedshu

Speak for yourself. The i5-3427U in my Macbook Air is plenty powerful and sips power while doing it.