$64 Pentium G4560 is neck and neck against i5-2500K

[Energycore]

1 hour ago, Kloaked said:

 

I saw Tweaktown in the source.

I've yet to experience tweaktown lying outright, but considering they go out of their way to show you a bunch of ads around adblock, I have decided to restrain myself from ever clicking their stuff.

[vanished]

I'm not sure why I just thought of this, but I wonder what the profit is on an i7 like a 6700/7700k, vs an i3, like the 6100 for example.  The i7 costs about twice as much, as it should since it's literally twice the chip, but does the profit scale linearly?  Would Intel potentially have a reason to actually sell more i3s than i5s/i7s?  Based on my (admittedly very limited) understanding of how all the chips are made, in theory, every production run should create far more i3s than i5, and more i5s than i7s.  If, for example, i3s are effectively 1/4 the cost to produce and sell for 1/2 the cost, their profit on them would be much better, assuming they sell enough of the things.  Plus, it's probably that someone will try the i3, be impressed enough to enjoy it, but also be left longing for more, which will lead them to buy another chip (an i5 or i7), thus increasing profit further.

 

Just my conspiracy ramblings on why we seem to see i3s talked up so much lately... 

[MageTank]

1 minute ago, Ryan_Vickers said:

I'm not sure why I just thought of this, but I wonder what the profit is on an i7 like a 6700/7700k, vs an i3, like the 6100 for example.  The i7 costs about twice as much, as it should since it's literally twice the chip, but does the profit scale linearly?  Would Intel potentially have a reason to actually sell more i3s than i5s/i7s?  Based on my (admittedly very limited) understanding of how all the chips are made, in theory, every production run should create far more i3s than i5, and more i5s than i7s.  If, for example, i3s are effectively 1/4 the cost to produce and sell for 1/2 the cost, their profit on them would be much better, assuming they sell enough of the things.  Plus, it's probably that someone will try the i3, be impressed enough to enjoy it, but also be left longing for more, which will lead them to buy another chip (an i5 or i7), thus increasing profit further.

 

Just my conspiracy ramblings on why we seem to see i3s talked up so much lately... 

I think the reasons i7's are more expensive has less to do with their relative performance (linear or not) to the i3's and i5's, and more so to do with the fact that the i5's and i3's are likely cutdown from the i7's to begin with. Since the yields for the i7's are likely not going to be high, they have to do something with the chips that failed to be i7's. Take the failed chip, clock it lower, turn HT off, boom, you have an i5. The same process can be repeated for i3's from i5's. I am no more educated on the subject than you are, so this is just my guess as well. 

 

@patrickjp93 would know, but the man has been a ghost lately. Am actually missing him, lol. 

[Guest Kloaked]

6 minutes ago, Energycore said:

I've yet to experience tweaktown lying outright, but considering they go out of their way to show you a bunch of ads around adblock, I have decided to restrain myself from ever clicking their stuff.

They take part in the rumor spreading and they have often quoted other websites as their sources.

 

The bad thing about that second part is those websites also quote Tweaktown as their source sometimes. Neither of them ever have an actual source.

[Energycore]

1 minute ago, Kloaked said:

They take part in the rumor spreading and they have often quoted other websites as their sources.

 

The bad thing about that second part is those websites also quote Tweaktown as their source sometimes. Neither of them ever have an actual source.

Lol good ole circular references. Yeah any bit of news that is just a copy of the source is basically useless. Journalism is more than this dammit!

[vanished]

2 minutes ago, MageTank said:

I think the reasons i7's are more expensive has less to do with their relative performance (linear or not) to the i3's and i5's, and more so to do with the fact that the i5's and i3's are likely cutdown from the i7's to begin with. Since the yields for the i7's are likely not going to be high, they have to do something with the chips that failed to be i7's. Take the failed chip, clock it lower, turn HT off, boom, you have an i5. The same process can be repeated for i3's from i5's. I am no more educated on the subject than you are, so this is just my guess as well. 

Yes, that's my understanding too.  They print a big disk of dies, and the ones in the middle are the purest, degrading more toward the edge, like so:

 

If this is correct, I think my theory is right, since there should be way more i3s, and yet they sell for ~half the price of a (theoretically much rarer) i7

[MageTank]

2 minutes ago, Kloaked said:

They take part in the rumor spreading and they have often quoted other websites as their sources.

 

The bad thing about that second part is those websites also quote Tweaktown as their source sometimes. Neither of them ever have an actual source.

Or in this case, you get a source, but the source itself has incomplete information. We don't even know what kind of hardware was used to test the other many CPU's they listed in this test. No mention of motherboards, memory speeds, GPU's (and their relative clock speeds). We did see a test mention two SLI 980 Ti's, so I assume they used a single 980 Ti in most tests? Either way, half-baked information like this isn't exactly helpful if it raises more questions than answers. 

[ARikozuM]

@Ryan_Vickers From my understanding the chips are made according to the overall best CPU they try to achieve and then test the wafer for stability. Once the wafer passes, they cut the dies and test the individuals for potential performance and grade them according to batches, possibly explaining why we only have about 12 CPUs this generation for desktop whereas the [4XXX series] feels like it had 30 or so. 

[Drak3]

@MageTank @Ryan_Vickers

 

While salvaging some failed i7's is used to make i5's, i3's, Pentiums, etc. Not all of the lower tier components are going to be failed i7's. Demand also plays a role on how many CPUs become what they are.

 

Say out of 100 people, 10 want i7's, and 90 want i5's (and can't afford to go higher). If Intel made 100 CPUs, and 50 of them qualify as i7's, Intel will still take the small hit in profit per unit, and take validated i7's and knock them down a peg. They'd make more money that way than just having 40 i7's sitting around for consumers that don't exist.

Rinse and repeat for people that can't afford the i5. Then the i3. Then the Pentium.

 

43 minutes ago, MageTank said:

@patrickjp93 would know, but the man has been a ghost lately. Am actually missing him, lol. 

I'm glad he hasn't been around. Although he's been replaced by rrubberr.

[Zodiark1593]

1 hour ago, MageTank said:

I think the reasons i7's are more expensive has less to do with their relative performance (linear or not) to the i3's and i5's, and more so to do with the fact that the i5's and i3's are likely cutdown from the i7's to begin with. Since the yields for the i7's are likely not going to be high, they have to do something with the chips that failed to be i7's. Take the failed chip, clock it lower, turn HT off, boom, you have an i5. The same process can be repeated for i3's from i5's. I am no more educated on the subject than you are, so this is just my guess as well. 

 

@patrickjp93 would know, but the man has been a ghost lately. Am actually missing him, lol. 

I3s are fully enabled, native dual core chips (though the Pentium and Calero lines come from cut i3 dies).  The quad core is a separate die.

 

[vanished]

13 minutes ago, Zodiark1593 said:

I3s are fully enabled, native dual core chips (though the Pentium and Calero lines come from cut i3 dies).  The quad core is a separate die.

 

So how would you redraw the wafer above?  It sounds like there would be more than 1 for starters.

[JoostinOnline]

3 hours ago, 8-Bit Ninja said:

in games that run fine on 2 cores im not surprised, but given the fact that when far cry primal was released it wouldn't even open a dual core. I would still take the 2500k with 4 cores 

New games can run just fine on dual core CPU's.  Old ones can run even better, because they tend to have better single-thread performance.  Far Cry games actually CAN run fine on dual core.  You just have to get past the dev's idiotic choice to lock it.

[ARikozuM]

4 minutes ago, JoostinOnline said:

New games can run just fine on dual core CPU's.  Old ones can run even better, because they tend to have better single-thread performance.  Far Cry games actually CAN run fine on dual core.  You just have to get past the dev's idiotic choice to lock it.

LTT Rules

#1 - Dual-cores are worthless

#2 - They get locked out of games and won't hit any good frame-rate that is above 240Hz TruMotion

#3 - Get the i7 from 20 years ago over the i5 from yesterday.

 

I'm completely kidding by the way, but that's how CPU threads go most of the time.

[Drak3]

3 minutes ago, JoostinOnline said:

New games can run just fine on dual core CPU's.  Old ones can run even better, because they tend to have better single-thread performance.

A dual core has the same IPC as a quad core of the same generation, meaning that if it runs well on a dual core, it'll run no worse on a quad core of same clock speeds. The only ways a dual core can outperform a quad core are:

A) The quad core has a lower clock speed, and the games don't take advantage of more than 2 cores.

B) The quad core is significantly older, resulting in lower IPC.

 

New games are also leveraging quad core CPUs more and more. Dual cores are becoming restrictive for the ambitious of AAA developers.

Honestly, dual cores should be limited to office and basic use scenarios at this point. They just aren't worth the effort and compromises anymore.

[JoostinOnline]

7 minutes ago, Drak3 said:

A dual core has the same IPC as a quad core of the same generation, meaning that if it runs well on a dual core, it'll run no worse on a quad core of same clock speeds. The only ways a dual core can outperform a quad core are:

A) The quad core has a lower clock speed, and the games don't take advantage of more than 2 cores.

B) The quad core is significantly older, resulting in lower IPC.

 

New games are also leveraging quad core CPUs more and more. Dual cores are becoming restrictive for the ambitious of AAA developers.

Honestly, dual cores should be limited to office and basic use scenarios at this point. They just aren't worth the effort and compromises anymore.

Take a look at single thread benchmarks sometime.  i3's tend to run better.  And it's not uncommon to see an i3 at a higher clock-speed.

 

Edit: Also, what games are you talking about besides a couple Far Cry ones?

[Zodiark1593]

8 minutes ago, Ryan_Vickers said:

So how would you redraw the wafer above?  It sounds like there would be more than 1 for starters.

For the consumer line, I believe there were 4 different dies total (as of Skylake). 2 core with GT2 igpu (i3, Pentium, Celeron), 4 core with GT2 (i5, i7, Xeon), 2 core with GT3, and 4 core with GT3 (the latter two for your xxxxHQ chips). 

 

The dies with the GT2 gpu are far more common as they are substantially smaller than the GT3 dies. The premium mobile space is where the GT3 dies are more prevelant (Macbooks). 

 

It is the server parts where cores get cut down from the base design (LGA 2011). 

[AresKrieger]

2 hours ago, Energycore said:

I've yet to experience tweaktown lying outright, but considering they go out of their way to show you a bunch of ads around adblock, I have decided to restrain myself from ever clicking their stuff.

Also to add to the circular reporting invalidity the fact that they haven't outright lied doesn't really mean anything when you consider these tests don't have any context

 

Just look a CNN's coverage of a women telling people to burn the suburbs rather than the urban area, they twisted that into her simply saying not to destroy the urban area via omission.

[79wjd]

18 minutes ago, Ryan_Vickers said:

So how would you redraw the wafer above?  It sounds like there would be more than 1 for starters.

My guess would be that yields aren't terrible and as such they make an i7 wafer that they artificially create i5s from saved create i3s and i5s from the defects. Then they also have i3 wafers to result in enough i3s since I would assume it's far more economical (unless yields are awful) to produce two different wafers. 

 

Although the one thing that makes me question that is that the iGPU makes up the majority of the die AND arent all CPU dies the same size?

[Energycore]

4 minutes ago, djdwosk97 said:

Although the one thing that makes me question that is that the iGPU makes up the majority of the die AND arent all CPU dies the same size?

I bet dies that have defects in the iGPU become no-iGPU Xeons

[Drak3]

1 minute ago, JoostinOnline said:

Take a look at single thread benchmarks sometime.

And find what?

At the same clock speeds, the same results as an i5 and i7 (within margin of error).

Hell, the single thread difference between the i3-7350K and i5-6600K is directly tied to the clock speeds of the chips. Same with the 7350K and 7600K. The 7350K and 7700K have the same single thread performance, and they rock the same clock speed.

[Princess Luna]

6 minutes ago, Drak3 said:

And find what?

At the same clock speeds, the same results as an i5 and i7 (within margin of error).

Hell, the single thread difference between the i3-7350K and i5-6600K is directly tied to the clock speeds of the chips. Same with the 7350K and 7600K. The 7350K and 7700K have the same single thread performance, and they rock the same clock speed.

Well for what is worth games are starting to multi-thread better nowadays, The Witcher 3 for instance takes good advantage of multi-threading this might keep something like the Hyper-Threading of the 7700 more attractive than the higher clock speed of an overclocked 7600k... or at least I believe so...

[JoostinOnline]

1 minute ago, Drak3 said:

And find what?

At the same clock speeds, the same results as an i5 and i7 (within margin of error).

Hell, the single thread difference between the i3-7350K and i5-6600K is directly tied to the clock speeds of the chips. Same with the 7350K and 7600K. The 7350K and 7700K have the same single thread performance, and they rock the same clock speed.

I hit Submit Reply too soon.  I edited it (twice actually), but I guess you'd already started responding.  i3's usually run at higher clock speeds than i5's, and therefore have higher single thread performance.  With hyper-threading, they can do some decent multi-thread processing.  Not as good as most quad-cores, but still plenty if you're just doing mid-range gaming.

 

Far Cry 4 ran perfectly fine on dual core CPU's once you got past their ridiculous block.  The recommended i5 would be crushed by any modern i3, both in single and multi-thread performance.  As I recall, they were mostly just using one core anyway.

 

But other than those games, which ones are (falsely) saying they require a quad core to run?

[ARikozuM]

11 minutes ago, djdwosk97 said:

My guess would be that yields aren't terrible and as such they make an i7 wafer that they artificially create i5s from saved create i3s and i5s from the defects. Then they also have i3 wafers to result in enough i3s since I would assume it's far more economical (unless yields are awful) to produce two different wafers. 

 

Although the one thing that makes me question that is that the iGPU makes up the majority of the die AND arent all CPU dies the same size?

Maybe they're an multi-chip module. I don't see why Intel would necessarily make an iGPU directly attached to a CPU when they could get more inventory by separating them.

[Drak3]

1 minute ago, Princess Cadence said:

Well for what is worth games are starting to multi-thread better nowadays, The Witcher 3 for instance takes good advantage of multi-threading this might keep something like the Hyper-Threading of the 7700 more attractive than the higher clock sleep of an overclocked 7600k... or at least I believe so...

That's my point. Games are taking advantage of the resources of more powerful chips. I believe GTA V and City Skylines both can leverage chips like the 5820K/6800K, although the gains are piss poor, and not worth spending the price delta of an x99 board.

 

JoostinOnline is making the claim that dual cores are still hanging with quad cores in terms of gaming performance.

 

4 minutes ago, JoostinOnline said:

Far Cry 4 ran perfectly fine on dual core CPU's once you got past their ridiculous block.

When the game first came out, no i3 machine could run the game and be called a decent performer/ acceptable experience, regardless of settings used. It was Ubisoft going back in and reworking it that caused it to run acceptably, but it wasn't without headache on players with machines that could run the game beforehand because of buggy updates, on a game by a developer that doesn't hold a track record of stellar PC ports.

 

It was a wasted effort to appease a minority that is holding some performance aspects of PC gaming back.

 

8 minutes ago, JoostinOnline said:

But other than those games, which ones are (falsely) saying they require a quad core to run?

There is a difference between running as intended, and running at all. GTA V runs on an i3, but it's a piss poor experience compared to an i5 or i7. Developers also acknowledge this, and that's why they are recommending Core i5's for a good experience with the games.

[ARikozuM]

10 minutes ago, Drak3 said:

That's my point. Games are taking advantage of the resources of more powerful chips. I believe GTA V and City Skylines both can leverage chips like the 5820K/6800K, although the gains are piss poor, and not worth spending the price delta of an x99 board.

Cites: Skylines only uses 3 cores... Clock speed and IPC make the most difference for this game.