We need some tips for writing quality reviews to assist newbies to the site and newbies to our area.

Jon

A CPU processor review should include its system chipset support and compatibility on 1) Motherboard and 2) RAM. Since these two should work together hand in hand, a mention of their frequency and synch operating speed should be of useful to people who plan to upgrade their PC or build a new one (if the motherboard supports the CPU they plan to buy). Therefore, a processor review might entail something along this line:

Clocking speed
Support of system bus – Front Side Bus (FSB)
Memory buffer (Cache, its processing level and memory handling)
Chipset compatibility
Architecture (pin socket, design type,Hyper-Threading, …)

A discussion of its performance may include:

- core operating speed
- clocking capability (in synch with RAM)
- management of heat
- its overclock capabilities
- overclock method – whether via motherboard’s BIOS or software; via FSB or multiplier, or a combination of both
- report of benchmark might be useful for some general information
- system performance as a whole (might also help to include both normal and overclock)

I think a brief discussion of the CPU architecture; its technology and family might be helpful. It also helps to include some discussion on installation, how the heatsink and fan assemble or work, etc. I think the most important thing in a processor review is the inclusion of its chipset compatibility, how the system works together with motherboard, RAM and CPU.

Thanks for your post, it is really appreciated.

Jon

I think all processor reviews should have a benchmark reported. I know this will fly in the face of several folks who think that we are asking too much from a review, but it is relatively easy to download SisSoft Sandra 2004 (see website below), run a quick test and report said findings.

http://www.sisoftware.co.uk/

The reason I say this is important is that to the average consumer there is no visible difference between the chips. They look the same, they install the same, and they do the same functions. However, the chips are significantly different in capability, based on FSB, cache and clock speed. In order to quantify this difference, you have to use a benchmark-type program, or else you are left with statements like "It feels faster than the other chip with the same clock and slower FSB"...

I do agree with Paul that along with this information, the big three are cache size, front side bus and clock speed. For those that do overclock, please explain what the chip can do and how you did it. This helps the subset of users who are looking for this type of information.

I personally think that this will garner a VH, and the rest is just gravy. If you want to include architecture, road maps to the future, what size process the chip is produced on, how many transistors are on the chip, etc., go ahead. This is not necessary, and a search of the manufacturers homepage can divulge all of this for those interested. If you feel obliged, leave the link to the manufacturer's website; it can't hurt!

Jason

I can't agree that posting a benchmark should be a requirement or that it is even that helpful. As benchmarks for CPU's are also affected by the interfaces and other components on a motherboard, they can be very misleading. With the number of different system boards available, the HSF's that can be used, supported hardware, and loading, CPU scores can differ significantly.

What I maintain is a generalization concerning performance with some comparison of speed, shown by past experience and/or the current experiences, as the system is used would be better for the average consumer ie. system builder. Including motherboard, software, hardware, and use information with this is more critical for average consumers and advanced builders. Arguably, a benchmark can be of benefit to some - but it is not critical in relating the experienced level of performance nor is it fully reliable due to the afore mentioned. "Overclockers" may appreciate the benchmark info but they represent a much smaller segment of the consumers who understandably will gain no benefit from numbers they cannot understand.

In short the only real benefit of providing a benchmark such as System Sandra would be in a "shootout" using the same processor in a number of different combinations which could then be compared to other processors in the same applications.

Providing links to other sites - this can be very distracting to consumers looking for a quick quality review that will influence their decision, and the web site use here at Epinions should be always be encouraged in preference to others. At times it is un-avoidable, but many times relevant product information can be copied directly from the manufacturers site if it is in open phrase form, sentences including the proper credits can also be provided. I know I don't want to explore another site for the pertinent information I want, as it should be available in the reviews I read.

A good discussion by the reviewer of the features provided by the manufacturer can easily be Very Helpful if not Most Helpful if a thorough job is done. Often these features are provided under "VIEW DETAILS" here at Epinions but can be found on the product literature and elsewhere. Any "green bar" ratings should also be explained. If terms are used, many will not understand, then some explanation of them should be included.

Alex

I resent your taking my words out of context here and elsewhere and I will remind you, you are addressing a fellow Advisor. I have not done this to you, but I can assure you I am very capable, however that is not proper behavior and I will not engage in it.

I will explain what you tried to distort only once.

Just one point is your seizing on the fact I often ask reviewers to check under "View Details" as I mentioned above. This is a good place, where details are provided, for reviewers to start when interested in knowing what to discuss.

Do you really want an in depth analysis of every word you write from a skewed viewpoint? Did you not understand the previous and adjacent sentence where I said; "A good discussion by the reviewer of the features provided by the manufacturer can easily be Very Helpful if not Most Helpful if a thorough job is done." Just what part of that do you not understand? It is completely self supporting.

Calling my words wishy-washy when the entire statement was quite clear...

Your portrayal of "feels faster" was nothing I said. Since when does the sense of touch have anything to do with electronic speed? Further, without adjectives how do you expect to relate thoughts? Without them, almost nothing can be conveyed. Benchmarks as applied to "home-built" systems do not provide a fair comparison due to all kinds of factors, and many will never understand them. I can understand where someone untrained in electronics would have difficulty with this. Benchmarks are no more than generalizations in reality and are not true indicators of performance - as the programmers that wrote these programs will admit themselves, and as there are too many unknowns to account for in every use. Placing a number - the manufacturers have already done this for us. This is why systems are built and then physically tested by OEM manufacturers. Those tests are stress related to catch design flaws and to provide longer product life. Some benchmarks do exactly this and can badly damage a system someone has worked hard to build that would never have failed in their use otherwise. You definitely have missed the point. If an individual is or isn't comfortable stressing his system, that choice should be his own and not decided for him because he wanted to write a review.

Alex,

I don't believe that I took anything out of context; I may have misinterpreted your meaning and therefore asked you for clarification, but nothing was taken out of context so far as I can see. I should make something clear right off the bat; I agree that you can rate how you see fit based on your own guidelines; I don't dispute that. In this case I thought we were discussing the merits or lack thereof of benchmarks in a processor review.

I feel that benchmarks are important to a processor review. You do not. As you initiated a discussion refuting the need for benchmarks following up on my post, I responded. Part of my response was asking you for clarification regarding how someone would describe the experienced level of performance without using metrics such as a benchmark. I never implied, inferred or took out of context the fact that you stated folks should use words such as 'feels faster'; rather I asked for clarification of how someone discusses experienced level of performance without only words such as this or quantification through benchmarks. I don't see how it is possible, and was asking for clarification.

I did not take out of context your statement regarding what makes a processor review VH or MH. If you look, I quoted that portion of your text word for word without slipping any of my own context in and then stated that in concept I agreed with the sentiment. I most certainly did not say that adjectives could not be used in a review, as that would be ridiculous. What may have happened was a misinterpretation of your meaning here and in comment sections around the site, and if so I apoligize.

One portion of your comment I do feel the need to ask you about, as I do not understand what your are driving at: