Six Core AMD reviewed

[El_Capitan]

Ok I'll try again.

 

Overclockers (like us members here) are only a small percentage of the readers. Most people do not overclock their CPUs. These are sold at a speed that is guaranteed to work. They want to know the performance they are getting out of the box. Of course if you tune the Bloomfield, it's going to beat it! But what tells you it is going to be easy? Nothing. You're making an assumption here. Everyone knows that every chip is different.

 

For manufacturers, increasing the frequency is just another way to increase performance, as adding more cores or adding more threads is. In Intel's case, yes, their architecture is more powerful for a given clock. So what? It's like saying that a Porshe will go faster than a firetruck with a given engine. You need to equip the firetruck with a more powerful engine if you want it to reach the same speed.

~Sigh.

 

Okay, you win. The 1090T wins as well. I'll go to bit-tech.net now.

[The Smith]

~Sigh.

 

Okay, you win. The 1090T wins as well. I'll go to bit-tech.net now.

I did not say a clock-for-clock comparison wasn't interesting! I just said this is not what most people want to see. Actually this might make for a good article, although we can already guess the answer; Intel yields better performance at the same clock.

[Bosco]

Personally I think thats a waste of time.

 

If you have two CPU's that are clocked differently, and the lower clocked CPU performs better what is the point in downclocking it the CPU that has the higher clock? To show a one sided victory?

 

We test how it its rated and overclock it from there, simple as that. Why people would care how things do clock for clock is beyond me. Those days are long gone, if you want something that compares clock for clock buy a slower CPU.

[El_Capitan]

I did not say a clock-for-clock comparison wasn't interesting! I just said this is not what most people want to see. Actually this might make for a good article, although we can already guess the answer; Intel yields better performance at the same clock.

I understand what you're saying. Yes, if I were a consumer who ran things at stock speeds, the 1090T is a good buy. We just have two different viewpoints.

 

I think yours is that most consumers for high-end CPU's aren't going to overclock it.

 

My thoughts are that most consumers for high-end CPU's are.

 

Why? They have Extreme and Black Editions for a reason, but at a higher price point. The TIM and CPU Cooling industries are doing quite well - look at all the different manufacturers. If no one overclocked (past an extra 400 MHz they could do on a stock heatsink), they would stick with the stock heatsinks.

 

Bit-tech.net, in my opinion, gave the best review for the 1090T. It compared stock CPU's as well as overclocked CPU's.

 

Now, I agree, there are a lot of people out there that don't overclock because there's still many that don't know how to. I just don't think those are the ones that are going to jump at buying the 1090T or the 980X.

Edited April 30, 2010 by El_Capitan

[El_Capitan]

Personally I think thats a waste of time.

 

If you have two CPU's that are clocked differently, and the lower clocked CPU performs better what is the point in downclocking it the CPU that has the higher clock? To show a one sided victory?

 

We test how it its rated and overclock it from there, simple as that. Why people would care how things do clock for clock is beyond me. Those days are long gone, if you want something that compares clock for clock buy a slower CPU.

We don't want to know how it really performs at the same clock speeds, but at the maximum stable 24/7 speeds for the lowest common denominator. Since the 1090T has the lowest overclocked frequency compared to the i7 930, both would be tested at 3.87 GHz, or 4.1 GHz... whatever the highest speed the 1090T can overclock to and still be stable. While another test to show how much more the i7 930 can be pushed to (4.4 GHz). It just gives an extra dimension to the price-performance of a high-end enthusiast CPU.

 

Why not do a poll? A 1090T review... test it at stock speeds, or at maxed overclock speeds?

 

Then include another poll for CPU's to test alongside with it, and whether at stock speeds or at maxed overclock speeds.

 

Look at it this way. As a consumer looking at SSD's, what is that consumer looking for? Access Speeds are relatively the same. The things I think are important are I/O speeds, and then sequential Read/Write speeds. There's also cost/GB, firmware, and stability. Then again... there will be those who ARE looking at getting I/O speed SSD's that want to know how well they perform in different RAID configurations. If it's cheaper to get two 64GB in RAID 0 than one 128GB, then most people are going to do it. It's the same with GPU's and comparing a single 5879 vs two 5770's in Crossfire or SLI. Cost per performance-wise, the two 5770's win.

Edited April 30, 2010 by El_Capitan

[cirro]

Most people do not overclock their CPUs. These are sold at a speed that is guaranteed to work. They want to know the performance they are getting out of the box. Of course if you tune the Bloomfield, it's going to beat it! But what tells you it is going to be easy? Nothing. You're making an assumption here.

 

[Bosco]

On the next cpu review in two weeks or so you will see the new graphs where it will show all the CPUs at stock speeds, then the next graph will have all overclocked scores side by side.

 

Hopefully that will give you some of the info you are looking for.

[redtigerdragon]

We don't want to know how it really performs at the same clock speeds, but at the maximum stable 24/7 speeds for the lowest common denominator. Since the 1090T has the lowest overclocked frequency compared to the i7 930, both would be tested at 3.87 GHz, or 4.1 GHz... whatever the highest speed the 1090T can overclock to and still be stable. While another test to show how much more the i7 930 can be pushed to (4.4 GHz). It just gives an extra dimension to the price-performance of a high-end enthusiast CPU.

 

This is already done...kinda. The charts show how the new CPU compares to others, both with it at stock speeds and OC'd. If you look at the original review for the CPU you want to pit it against, the same graphs will be there (assuming you aren't comparing too old of a CPU). Then, you can just compare the highest OC that the reviewer got on both CPU's (it will be listed under the OC section of the review). You can also compare how each CPU scored OC'd on each benchmark, using the two graphs.

 

I'll give you the point that it would be somewhat beneficial to not only show the main competition at stock, but also OC'd on the new charts of the new review.

 

^This was answered by Bosco just before i posted. STAY OUT OF MY MIND!! AHHHHH!

 

Let me know if this doesn't answer your question at all, but I think it's what you were hinting at...

Edited April 30, 2010 by redtigerdragon

[JBags]

IMO, clock-for-clock testing would be interesting, but only really relevant if both chips in the comparison had an equal number of cores/threads between different manufacturers (ie i7 980X w/o HT and Phenom ii x6) OR between newer chips of the same manufacturer, Phenom vs Phenom II, or C2Q vs i5/i7 with no HT. In the first case, it would reveal which architecture is better at what calculations/programs/benchmarks and in the second it would reveal the improvements over the older architecture.

 

to me, comparing clock-for-clock an i7 920/30 with or without HT against the Phenom II x6, or an 980X with HT against a Phenom II X6 would be pointless as they would not reveal anything of importance other than 6 cores beats 4 or 8/12 threads beats 6. well, yeah, duh.

 

 

El_Capitan, I do understand what you're saying, a 920/30 with HT enabled matches or exceeds the 6 core Phenom II at relatively equal clock speeds, while also being similarly priced. I can dig that.

 

However, that would require overclocking the i7. There are in fact, many people out there that just want a high-end chip, just to have an high-end chip, not to OC it. Money burns a hole in their pockets. So for them, the Phenom II X6 beats the 920 at stock, so that's what they'll get, or they'll get the 980X because, at stock, it beats the phenom ii x6. I personally know a few people like this, I'm sure everyone does.

 

But a lot of people here look for price/performance at stock (of course this ratio increases as one overclocks). I'm sure if you browsed the System Builds subforum, you'd see a lot of these type posts, people looking to get the most performance for their dollar. That's not to say these people don't overclock, nor that they do, either. But completely discarding stock clock comparisions in favor of clock-for-clock singles out these people while instead providing relatively useless numbers for the situations i named above. However! Those comparisions given against chips of equal cores/threads, would provide a cool figure if added to reviews.

 

 

But of all things! People do not always agree on everything all the time. And that is no reason so stop frequenting OCC, few people disagreed with you on one thing. OCC is a great place, and if you find bit-tech's reviews more relevant to you, then by all means, you're free to hold them with higher regard, but that's no reason to quit OCC...

[El_Capitan]

@JBags - I was just kidding that I would stop coming here. It's fun.

 

Besides, who wants to yak with a bunch of brits, anyway? (bit-tech.net)

[joel.monteiro]

on the next cpu review can u put Minimum frames plz i want to see those?

[Bosco]

Maybe down the road but for now we are sticking with the average.